Agricultural product data recommendation optimization analysis system and method based on artificial intelligence

Abstract

The invention discloses an agricultural product data recommendation optimization analysis system and method based on artificial intelligence, and relates to the technical field of agricultural production.The method comprises the steps that firstly, phenotypes of agricultural products are obtained from variety approval, and the phenotype collaboration or antagonism relation is judged according to a grey relational analysis method; establishing a data matrix, and calculating a weight through an entropy method to obtain a variety adaptability comprehensive evaluation index; acquiring soil data such as organic carbon acidity and alkalinity of soil in a planting area and meteorological data such as rainfall and accumulated temperature in a growth period; and constructing a variety recommendation model based on multiple linear regression and optimizing parameters. The system comprises a phenotype correlation analysis module, an adaptability evaluation module, an environmental data acquisition module and a variety recommendation modeling module which are connected in sequence. According to the scheme, the variety environment interaction relation is quantified through multi-source data fusion, varieties are accurately recommended, the introduction trial and error cost is reduced, and the planting economic benefits are improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural production technology, and specifically to an agricultural product data recommendation optimization analysis system and method based on artificial intelligence. Background Art

[0002] Annual precipitation in dryland areas generally falls below 400 mm, and its spatial and temporal distribution is uneven, leading to significant fluctuations in soil moisture. Under these conditions, the optimal selection of agricultural product varieties based on local environmental characteristics to ensure high yields, high quality, and reduced field management costs is a pressing issue. Variety adaptability requires comprehensive consideration of climate, soil, pests and diseases, and local cultivation and management characteristics. Considering only climate similarity often limits the scope of selection and fails to account for the impact of climate factors such as temperature, sunshine, and rainfall on agricultural adaptability. Experimenting with introduced varieties in local areas can also be costly and time-consuming. Therefore, selecting the most suitable varieties from a wide range of varieties based on multi-source data fusion analysis is key to improving agricultural production efficiency.

[0003] Currently, recommendations for agricultural product varieties are mostly made by local agricultural extension departments. Due to limited space, small sample sizes, and a lack of intelligent tools, the variety recommendations are limited and inaccurate. Farmers currently rely on experience, recommendations from close friends, or distributors rather than scientific evidence, making selection difficult. Some farmers single-mindedly pursue high yields, ignoring the relationship between local climate, soil conditions, and variety adaptability, resulting in low yields and poor quality. In the arid and semi-arid regions of Northwest China, in particular, the limited variety of locally cultivated agricultural products necessitates importation from other regions. Currently, some research uses random forest models to assess variety adaptability at the county level to provide decision support, but these approaches are limited to large environmental scales, and research on agricultural product variety recommendations is even rarer. Furthermore, the recent rise of big data technology offers significant advantages in processing massive, multidimensional data. It can be applied to analyze the characteristics of a large number of agricultural product varieties, as well as nationwide climate characteristics, to quantify the "variety-environment" interaction and accurately recommend suitable varieties.

[0004] In summary, existing research on agricultural product recommendations tailored to local climate characteristics, particularly those in arid climates, is incomplete and insufficient to meet the needs of agricultural practitioners. Therefore, developing precise variety recommendation models based on comprehensive analysis of comprehensive variety characteristics and large-scale environmental indicators to better guide agricultural production and promote high yields and quality is an urgent issue. Summary of the Invention

[0005] The purpose of the present invention is to provide an artificial intelligence-based agricultural product data recommendation optimization analysis method to solve the problems raised in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Step S1, obtaining phenotypes of different varieties of agricultural products, calculating correlation coefficients of different phenotypes by grey correlation analysis, and determining the correlation types between different phenotypes of agricultural products based on the correlation coefficients, wherein the correlation types include synergistic relationships and antagonistic relationships;

[0008] Step S2: calculating a comprehensive evaluation index of the variety adaptability of the agricultural product in the promotion area based on the correlation type between different phenotypes, wherein the promotion area is the area where the agricultural product is grown;

[0009] Step S3: Acquire soil data and meteorological data of the experimental planting area and recommended planting area of ​​the agricultural product;

[0010] Step S4: constructing a variety recommendation model based on the comprehensive evaluation index of variety adaptability, the soil data, and the meteorological data.

[0011] Furthermore, step S1 further includes the following steps:

[0012] Step S1-1: Obtain m phenotypes of n varieties of agricultural products from the seed industry big data platform, wherein the phenotypes include the growth period, plant characteristics, ear, yield, and resistance of the agricultural products; obtain data for each phenotype, and quantify the unquantified data using numerical labels; convert the phenotypic data into dimensionless values ​​according to the mean method:

[0013]

[0014] where x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, n represents the number of varieties, x i ` (k) represents the raw data of the kth phenotype of the i-th variety;

[0015] Step S1-2: setting an optimal value of the dimensionless value of each phenotype, and determining reference data of the phenotype according to the optimal value to form a reference data column;

[0016] x0(k)=[x0(1),x0(2),x0(3)...x0(m)]

[0017] Where x0(k) represents the reference data of the kth phenotype, and m represents the total number of phenotypes;

[0018] Calculate the absolute value of the difference between the dimensionless value of the phenotype of each variety of agricultural products and the dimensionless value of the corresponding phenotype in the reference data column, and compare to obtain the maximum and minimum absolute values ​​of the dimensionless value difference;

[0019] Step S1-3, calculating the association coefficient of the phenotype according to the absolute value of the difference:

[0020]

[0021] Where i represents the i-th variety, k represents the k-th phenotype, ξ i (k) represents the correlation coefficient of the kth phenotype of the i-th variety, x0(k) represents the reference data of the kth phenotype, and x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, ρ is the resolution coefficient and is set to 0.5, min represents the minimum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties, and maxmax represents the maximum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties;

[0022] Step S1-4, averaging the correlation coefficients of two different phenotypes of an agricultural product to obtain an arithmetic mean, and then averaging the arithmetic mean of the correlation coefficients of the two different phenotypes corresponding to each agricultural product to obtain a comprehensive average value; setting a comprehensive average value threshold, when the correlation coefficient is greater than or equal to the comprehensive average value threshold, the corresponding two different phenotypes are in a synergistic relationship; when the correlation coefficient is less than the comprehensive average value threshold, the corresponding two different phenotypes are in an antagonistic relationship.

[0023] Furthermore, step S2 further includes the following steps:

[0024] Step S2-1, obtaining promotion areas of agricultural products and establishing a data matrix, wherein the data matrix includes dimensionless values ​​of m phenotypes of n agricultural product varieties in the promotion areas;

[0025] Step S2-2: Divide the phenotypes into two categories: positive factors and negative factors and sort them; use the minimum value of the dimensionless value of the positive factors to normalize the data of the positive factors:

[0026]

[0027] Where i is the i-th agricultural product variety; j represents the order of the positive factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype;

[0028] Normalize the data for negative factors using the maximum dimensionless value of the negative factor:

[0029]

[0030] Where i is the i-th agricultural product variety; j represents the order of the negative factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype;

[0031] Step S2-3: Transform the data matrix and calculate the proportion of each phenotype:

[0032]

[0033] Among them, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, r ij represents the jth phenotype of the i-th agricultural product variety after data standardization, and m represents the total number of phenotypes;

[0034] Step S2-4: Calculate the entropy value and weight of each phenotype. The entropy value is calculated as follows:

[0035]

[0036] where e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, m represents the total number of phenotypes, j represents the jth phenotype of the i-th agricultural product variety, and ln represents the pair operation with base e;

[0037] The weights are calculated as follows:

[0038]

[0039] where w j represents the weight ratio of the jth phenotype of the i-th agricultural product variety, e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety;

[0040] Step S2-5: Calculate the comprehensive evaluation index of the variety adaptability of the agricultural product based on the weights and the elements in the data matrix:

[0041]

[0042] Among them VACEL i represents the comprehensive evaluation index of variety adaptability of the jth phenotype of the i-th agricultural product variety, w j represents the weight ratio of the jth phenotype, r ij represents the element represented by the jth phenotype of the i-th agricultural product variety in the data matrix, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety.

[0043] Furthermore, step S3 further includes the following steps:

[0044] Step S3-1, obtain a vector map of the agricultural product planting area, use a geographic information system to randomly mark points on the vector map, and obtain point coordinates according to a coordinate system;

[0045] Step S3-2: Acquire soil information for the experimental planting area, including organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, and soil thickness; and obtain meteorological data for the recommended planting area, including rainfall during the growing season, accumulated temperature, average daily minimum temperature, average temperature, and potential evapotranspiration.

[0046] Step S3-3: randomly select N blocks according to the coordinates of the point, obtain soil information and meteorological data of the N blocks, and form a data set {rainfall during the growing period, accumulated temperature, average daily minimum temperature, average temperature, potential evapotranspiration, soil organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, soil thickness}. The elements in the data set are environmental characteristics, and the environmental characteristics are processed using Z-score standardization.

[0047] Furthermore, step S4 further includes the following steps:

[0048] Step S4-1: construct a variety recommendation model using multiple linear regression. The variety recommendation model is:

[0049]

[0050] in, represents the comprehensive evaluation index of variety adaptability; w is the weight vector, each w j Represents the jth environmental feature x j Forecast results The influence of; b is the bias term, which is a constant term, indicating that when all features x i The output of the model when is zero; the values ​​of w and b are initially set randomly;

[0051] Step S4-2: Calculate the weight vector w according to the loss function method:

[0052]

[0053] Where L(w, b) represents the deviation between the predicted value of w and b and the actual value. The smaller the value, the closer the model prediction is to the actual adaptability. N represents N blocks, W T Represents the rank of the weight vector; y i represents the comprehensive evaluation index of variety adaptability of the i-th phenotype;

[0054] Step S4-3, calculate the partial derivative of the loss function with respect to the weight:

[0055]

[0056] where w jrepresents the impact of the jth environmental feature on the prediction result, N represents the number of blocks, x ij represents the jth environmental feature of the i-th block, where i is not greater than N;

[0057] Update the weight w according to the gradient descent method j ;

[0058] Step S4-4: setting a threshold value for the comprehensive evaluation index of variety adaptability; agricultural products whose comprehensive evaluation index of variety adaptability is greater than the threshold value are considered suitable for promotion.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. This invention uses grey correlation analysis to determine the synergistic or antagonistic relationship between agricultural product phenotypes, combines it with the entropy method to calculate phenotypic weights, and constructs a comprehensive evaluation index for variety adaptability. This comprehensively quantifies the interactions between variety characteristics, avoids the one-sidedness of single-indicator evaluation, improves the scientificity and comprehensiveness of variety adaptability assessment, and lays the foundation for accurate recommendations.

[0061] 2. The present invention integrates the soil data and meteorological data of the experimental planting area and the recommended planting area, uses multivariate linear regression to construct a variety recommendation model, and optimizes the model parameters with the help of the gradient descent method, breaking through the limitations of traditional experience-based recommendations, reducing the cost and time consumption of trial and error in introducing new varieties, and improving the accuracy and practicality of the recommendation results.

[0062] 3. The present invention obtains the coordinates of planting area points through a geographic information system and combines multi-source environmental data to construct a data set containing multi-dimensional environmental characteristics, so that the recommendation model can take into account the climate and soil conditions of different regions, especially for the problem of uneven precipitation in dry farming areas. It can accurately match suitable varieties and help achieve high yield, high quality and efficient management of agricultural products in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of an artificial intelligence-based agricultural product data recommendation and optimization analysis method of the present invention;

[0064] Figure 2 This is a structural diagram of an agricultural product data recommendation and optimization analysis system based on artificial intelligence in the present invention. DETAILED DESCRIPTION

[0065] 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.

[0066] Example 1: Figure 1 As shown, the present invention provides a technical solution, a method for optimizing and analyzing agricultural product data recommendation based on artificial intelligence, which includes the following steps:

[0067] Step S1, obtaining phenotypes of different varieties of agricultural products, calculating correlation coefficients of different phenotypes by grey correlation analysis, and determining the correlation types between different phenotypes of agricultural products based on the correlation coefficients, wherein the correlation types include synergistic relationships and antagonistic relationships;

[0068] Step S2: calculating a comprehensive evaluation index of the variety adaptability of the agricultural product in the promotion area based on the correlation type between different phenotypes, wherein the promotion area is the area where the agricultural product is grown;

[0069] Step S3: Acquire soil data and meteorological data of the experimental planting area and recommended planting area of ​​the agricultural product;

[0070] Step S4: constructing a variety recommendation model based on the comprehensive evaluation index of variety adaptability, the soil data, and the meteorological data.

[0071] Furthermore, step S1 further includes the following steps:

[0072] Step S1-1: Obtain m phenotypes of n varieties of agricultural products from the seed industry big data platform, wherein the phenotypes include the growth period, plant characteristics, ear, yield, and resistance of the agricultural products; obtain data for each phenotype, and quantify the unquantified data using numerical labels; convert the phenotypic data into dimensionless values ​​according to the mean method:

[0073]

[0074] where x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, n represents the number of varieties, x i ` (k) represents the raw data of the kth phenotype of the i-th variety;

[0075] Step S1-2: setting the optimal dimensionless value of each phenotype, such as the dimensionless value corresponding to the phenotype with the highest yield, the shortest growth period, the shortest plant, and the strongest disease resistance; determining the reference data of the phenotype based on the optimal value to form a reference data column;

[0076] x0(k)=[x0(1),x0(2),x0(3)...x0(m)]

[0077] Where x0(k) represents the reference data of the kth phenotype, and m represents the total number of phenotypes;

[0078] Calculate the absolute value of the difference between the dimensionless value of the phenotype of each variety of agricultural products and the dimensionless value of the corresponding phenotype in the reference data column, and compare to obtain the maximum and minimum absolute values ​​of the dimensionless value difference;

[0079] Step S1-3, calculating the association coefficient of the phenotype according to the absolute value of the difference:

[0080]

[0081] Where i represents the i-th variety, k represents the k-th phenotype, ξ i (k) represents the correlation coefficient of the kth phenotype of the i-th variety, x0(k) represents the reference data of the kth phenotype, and x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, ρ is the resolution coefficient and is set to 0.5, min represents the minimum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties, and maxmax represents the maximum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties;

[0082] Step S1-4, averaging the correlation coefficients of two different phenotypes of an agricultural product to obtain an arithmetic mean, and then averaging the arithmetic mean of the correlation coefficients of the two different phenotypes corresponding to each agricultural product to obtain a comprehensive average value; setting a comprehensive average value threshold, when the correlation coefficient is greater than or equal to the comprehensive average value threshold, the corresponding two different phenotypes are in a synergistic relationship; when the correlation coefficient is less than the comprehensive average value threshold, the corresponding two different phenotypes are in an antagonistic relationship.

[0083] Furthermore, step S2 further includes the following steps:

[0084] Step S2-1, obtaining promotion areas of agricultural products and establishing a data matrix, wherein the data matrix includes dimensionless values ​​of m phenotypes of n agricultural product varieties in the promotion areas;

[0085] Step S2-2: Divide the phenotypes into two categories: positive factors and negative factors and sort them; use the minimum value of the dimensionless value of the positive factors to normalize the data of the positive factors:

[0086]

[0087] Where i is the i-th agricultural product variety; j represents the order of the positive factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype;

[0088] Normalize the data for negative factors using the maximum dimensionless value of the negative factor:

[0089]

[0090] Where i is the i-th agricultural product variety; j represents the order of the negative factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype;

[0091] Step S2-3: Transform the data matrix and calculate the proportion of each phenotype:

[0092]

[0093] Among them, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, r ij represents the jth phenotype of the i-th agricultural product variety after data standardization, and m represents the total number of phenotypes;

[0094] Step S2-4: Calculate the entropy value and weight of each phenotype. The entropy value is calculated as follows:

[0095]

[0096] where e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, m represents the total number of phenotypes, j represents the jth phenotype of the i-th agricultural product variety, and ln represents the pair operation with base e;

[0097] The weights are calculated as follows:

[0098]

[0099] where w j represents the weight ratio of the jth phenotype of the i-th agricultural product variety, e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety;

[0100] Step S2-5: Calculate the comprehensive evaluation index of the variety adaptability of the agricultural product based on the weights and the elements in the data matrix:

[0101]

[0102] Among them VACEL i represents the comprehensive evaluation index of variety adaptability of the jth phenotype of the i-th agricultural product variety, w j represents the weight ratio of the jth phenotype, r ijrepresents the element represented by the jth phenotype of the i-th agricultural product variety in the data matrix, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety.

[0103] Furthermore, step S3 further includes the following steps:

[0104] Step S3-1, obtain a vector map of the agricultural product planting area, use a geographic information system to randomly mark points on the vector map, and obtain point coordinates according to a coordinate system;

[0105] Step S3-2: Acquire soil information for the experimental planting area, including organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, and soil thickness; and obtain meteorological data for the recommended planting area, including rainfall during the growing season, accumulated temperature, average daily minimum temperature, average temperature, and potential evapotranspiration.

[0106] Step S3-3: randomly select N blocks according to the coordinates of the point, obtain soil information and meteorological data of the N blocks, and form a data set {rainfall during the growing period, accumulated temperature, average daily minimum temperature, average temperature, potential evapotranspiration, soil organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, soil thickness}. The elements in the data set are environmental characteristics, and the environmental characteristics are processed using Z-score standardization.

[0107] Furthermore, step S4 further includes the following steps:

[0108] Step S4-1: construct a variety recommendation model using multiple linear regression. The variety recommendation model is:

[0109]

[0110] in, represents the comprehensive evaluation index of variety adaptability; w is the weight vector, each w j Represents the jth environmental feature x j Forecast results The influence of; b is the bias term, which is a constant term, indicating that when all features x i The output of the model when is zero; the values ​​of w and b are initially set randomly;

[0111] Step S4-2: Calculate the weight vector w according to the loss function method:

[0112]

[0113] Where L(w, b) represents the deviation between the predicted value of w and b and the actual value. The smaller the value, the closer the model prediction is to the actual adaptability. N represents N blocks, W TRepresents the rank of the weight vector; y i represents the comprehensive evaluation index of variety adaptability of the i-th phenotype;

[0114] Step S4-3, calculate the partial derivative of the loss function with respect to the weight:

[0115]

[0116] where w j represents the impact of the jth environmental feature on the prediction result, N represents the number of blocks, x ij represents the jth environmental feature of the i-th block, where i is not greater than N;

[0117] Update the weight w according to the gradient descent method j ;

[0118] Step S4-4: setting a threshold value for the comprehensive evaluation index of variety adaptability; agricultural products whose comprehensive evaluation index of variety adaptability is greater than the threshold value are considered suitable for promotion.

[0119] For example:

[0120] Fresh corn was selected as an example agricultural product, and three fresh corn varieties were selected from the "China Seed Industry Big Data Platform" to monitor three phenotypes: plant height (cm), ear length (cm), and ear diameter (cm).

[0121]

[0122] Convert the phenotype to a dimensionless value using the mean method:

[0123] Dimensionless value of plant height of variety A:

[0124]

[0125] Dimensionless value of ear length:

[0126]

[0127] Dimensionless value of ear diameter:

[0128]

[0129] The negative sign only indicates the direction of deviation from the mean.

[0130] The optimal value of each phenotype is taken as the reference data column, such as plant height 210 cm, ear length 25 cm, and ear diameter 5.0 cm, and the corresponding dimensionless values ​​of the three are 0.5, 1, and 1;

[0131] Calculate the absolute difference:

[0132] For example, the absolute difference between variety A and the reference column:

[0133] Plant height: |0.5-0.5|=0

[0134] Spike length: |0-1|=1

[0135] Ear diameter: |-0.5-1|=1.5

[0136] Similarly, calculate the absolute differences between variety B and the reference column: plant height 0.5, ear length 0.5, ear diameter 0.5;

[0137] The absolute differences between variety C and the reference column are: plant height 1, ear length 1.5, and ear diameter 1;

[0138] It can be seen that the global minimum value is 0 and the global maximum value is 1.5;

[0139] Take ρ = 0.5 to calculate the correlation coefficient. For example, the correlation coefficient of plant height of variety A is:

[0140]

[0141] The correlation coefficient of ear length of variety A is:

[0142]

[0143] The arithmetic mean of the correlation coefficient between plant height and ear length of variety A is: 0.7145

[0144] Similarly, the arithmetic mean of the correlation coefficient between plant height and ear length of variety B is: 0.6

[0145] The arithmetic mean of the correlation coefficient between plant height and ear length of variety C is: 0.381

[0146] The combined average of plant height and ear length is: 0.565

[0147] The threshold of the comprehensive mean value is set to 0.5. It is known that 0.565 is greater than 0.5, so there is a synergistic relationship between plant height and ear length;

[0148] Taking a province as a unit, a phenotypic data matrix containing three varieties was established, in which the positive factors were plant height and ear diameter, and the negative factor was ear length.

[0149] Positive factors are normalized by their minimum value, for example:

[0150] Standard plant height value of variety A:

[0151]

[0152] Negative factors are normalized with the maximum value, e.g.

[0153] Standard value of ear length of variety A:

[0154]

[0155] Let the standard value of ear length be 0.8, and establish the standardized data matrix of variety A:

[0156] r = [1, 0.5, 0.8]

[0157] Calculate the specific gravity, for example, the specific gravity of plant height is:

[0158]

[0159] Calculate the entropy of plant height:

[0160]

[0161] Let the entropy values ​​of other phenotypes be 0.8 and 0.7, then the weight of plant height is:

[0162]

[0163] The comprehensive evaluation index of variety adaptability is:

[0164] VACEL=0.25×1+0.3×0.5+0.45×0.8=0.76

[0165] The soil data obtained are: organic carbon 15g / kg, pH=7.2, total nitrogen 1.0g / kg, soil bulk density 1.2g / cm 3 .

[0166] The meteorological data obtained are: rainfall during the growing period is 280 mm, accumulated temperature is 2600 °C, daily average minimum temperature is 14 °C, and potential evapotranspiration is 500 mm.

[0167] Using multiple linear regression, the model expression is:

[0168] y = w1 × rainfall + w2 × accumulated temperature + w3 × soil organic carbon + b

[0169] Optimize parameters by gradient descent: Assuming the initial weights:

[0170] w = [0.1, 0.2, 0.3]

[0171] After iterative calculation of the loss function, the final weight is obtained:

[0172] w = [0.3, 0.5, 0.2]

[0173] Bias term:

[0174] b=0.1

[0175] Prediction results: Adaptability index of a certain variety in the region:

[0176] y=0.3×280+0.5×2600+0.2×15+0.1×1486.1

[0177] Determine whether the variety is suitable for promotion based on the threshold.

[0178] Example 2: Figure 2 As shown, the present invention provides an artificial intelligence-based agricultural product data recommendation optimization analysis system, which includes: a phenotypic association analysis module, an adaptability evaluation module, an environmental data acquisition module, and a variety recommendation modeling module; the phenotypic association analysis module is used to analyze the synergistic or antagonistic relationship between agricultural product phenotypes; the adaptability evaluation module is used to calculate the comprehensive evaluation index of agricultural product variety adaptability based on the synergistic or antagonistic relationship; the environmental data acquisition module is used to obtain soil data and meteorological data from agricultural product experimental planting areas and recommended planting areas to establish a data set; the variety recommendation modeling module is used to construct a variety recommendation model;

[0179] The output end of the phenotypic association analysis module is electrically connected to the input end of the adaptability evaluation module, the output end of the adaptability evaluation module is electrically connected to the input end of the environmental data acquisition module, and the output end of the environmental data acquisition module is electrically connected to the input end of the variety recommendation modeling module.

[0180] The phenotypic association analysis module includes a phenotypic data acquisition unit and an association relationship analysis unit; the phenotypic data acquisition unit is used to obtain the phenotypes of agricultural products from public variety approvals and convert the phenotypes into dimensionless interval values ​​through the mean method; the association relationship analysis unit is used to determine the reference data column based on the dimensionless values, calculate the absolute value of the difference and the association coefficient, and judge the synergistic or antagonistic relationship of the phenotypes.

[0181] The adaptability evaluation module includes a data matrix establishment unit and an evaluation index calculation unit; the data matrix establishment unit is used to obtain suitable promotion areas for agricultural products on a provincial basis and establish a data matrix containing the phenotypes of agricultural product varieties in each province; the evaluation index calculation unit is used to standardize the phenotype data and calculate the comprehensive evaluation index of variety adaptability based on the proportion, entropy value and weight of the phenotype.

[0182] The environmental data acquisition module includes a geographic point acquisition unit and an environmental data acquisition unit; the geographic point acquisition unit is used to obtain a vector map of the agricultural product planting area, randomly mark points using a geographic information system and obtain point coordinates; the environmental data acquisition unit is used to obtain soil information and meteorological data of the planting area to form a data set containing environmental characteristics.

[0183] The variety recommendation modeling module includes a model construction unit and a parameter optimization unit; the model construction unit is used to construct a variety recommendation model using multiple linear regression and set the initial weights and bias terms of the model; the parameter optimization unit is used to calculate the weight vector according to the loss function method and update the weight by the gradient descent method;

[0184] Finally, it should be noted that it will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be embraced therein. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

Claims

1. An artificial intelligence-based agricultural product data recommendation and optimization analysis method, characterized by: The method comprises the following steps: Step S1, obtaining phenotypes of different varieties of agricultural products, calculating correlation coefficients of different phenotypes by grey correlation analysis, and determining the correlation types between different phenotypes of agricultural products based on the correlation coefficients, wherein the correlation types include synergistic relationships and antagonistic relationships; Step S2: calculating a comprehensive evaluation index of the variety adaptability of the agricultural product in the promotion area based on the correlation type between different phenotypes, wherein the promotion area is the area where the agricultural product is grown; Step S3: Acquire soil data and meteorological data of the experimental planting area and recommended planting area of ​​the agricultural product; Step S4: constructing a variety recommendation model based on the comprehensive evaluation index of variety adaptability, the soil data, and the meteorological data.

2. The method for agricultural product data recommendation and optimization analysis based on artificial intelligence according to claim 1, characterized in that: The step S1 further includes: Step S1-1: Obtain m phenotypes of n varieties of agricultural products from the seed industry big data platform, wherein the phenotypes include the growth period, plant characteristics, ear, yield, and resistance of the agricultural products; obtain data for each phenotype, and quantify the unquantified data using numerical labels; convert the phenotypic data into dimensionless values ​​according to the mean method: where x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, n represents the number of varieties, x i ` (k) represents the raw data of the kth phenotype of the i-th variety; Step S1-2, setting an optimal value of the dimensionless value of each phenotype, and determining reference data of the phenotype according to the optimal value to form a reference data column; x0(k)=[x0(1),x0(2),x0(3)...x0(m)] Where x0(k) represents the reference data of the kth phenotype, and m represents the total number of phenotypes; Calculate the absolute value of the difference between the dimensionless value of the phenotype of each variety of agricultural products and the dimensionless value of the corresponding phenotype in the reference data column, and compare to obtain the maximum and minimum absolute values ​​of the dimensionless value difference; Step S1-3, calculating the association coefficient of the phenotype according to the absolute value of the difference: Where i represents the i-th variety, k represents the k-th phenotype, ξ i (k) represents the correlation coefficient of the kth phenotype of the i-th variety, x0(k) represents the reference data of the kth phenotype, and x i (k) represents the dimensionless value of the kth phenotype of the i-th variety, ρ is the resolution coefficient and is set to 0.5, min min represents the minimum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties, and max max represents the maximum absolute value of the difference between the dimensionless values ​​of all phenotypes of all varieties; Step S1-4, averaging the correlation coefficients of two different phenotypes of an agricultural product to obtain an arithmetic mean, and then averaging the arithmetic mean of the correlation coefficients of the two different phenotypes corresponding to each agricultural product to obtain a comprehensive average value; setting a comprehensive average value threshold, when the correlation coefficient is greater than or equal to the comprehensive average value threshold, the corresponding two different phenotypes are in a synergistic relationship; when the correlation coefficient is less than the comprehensive average value threshold, the corresponding two different phenotypes are in an antagonistic relationship.

3. The method for agricultural product data recommendation and optimization analysis based on artificial intelligence according to claim 1, characterized in that: The step S2 further includes: Step S2-1, obtaining promotion areas of agricultural products and establishing a data matrix, wherein the data matrix includes dimensionless values ​​of m phenotypes of n agricultural product varieties in the promotion areas; Step S2-2: Divide the phenotypes into two categories: positive factors and negative factors and sort them; use the minimum value of the dimensionless value of the positive factors to normalize the data of the positive factors: Where i is the i-th agricultural product variety; j represents the order of the positive factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype; Normalize the data for negative factors using the maximum dimensionless value of the negative factor: Where i is the i-th agricultural product variety; j represents the order of the negative factor in the ranking; r represents the element in the data matrix; x represents the dimensionless value of the phenotype; Step S2-3: Transform the data matrix and calculate the proportion of each phenotype: Among them, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, r ij represents the jth phenotype of the i-th agricultural product variety after data standardization, and m represents the total number of phenotypes; Step S2-4: Calculate the entropy value and weight of each phenotype. The entropy value is calculated as follows: where e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, P ij represents the proportion of the jth phenotype of the i-th agricultural product variety in the phenotypes of all agricultural product varieties, m represents the total number of phenotypes, j represents the jth phenotype of the i-th agricultural product variety, and ln represents the pair operation with base e; The weights are calculated as follows: where w j represents the weight ratio of the jth phenotype of the i-th agricultural product variety, e j represents the entropy value of the jth phenotype of the i-th agricultural product variety, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety; Step S2-5: Calculate the comprehensive evaluation index of the variety adaptability of the agricultural product based on the weights and the elements in the data matrix: Among them VACEL i represents the comprehensive evaluation index of variety adaptability of the jth phenotype of the i-th agricultural product variety, w j represents the weight ratio of the jth phenotype, r ij represents the element represented by the jth phenotype of the i-th agricultural product variety in the data matrix, m represents the total number of phenotypes, and j represents the jth phenotype of the i-th agricultural product variety.

4. The method for agricultural product data recommendation and optimization analysis based on artificial intelligence according to claim 1, characterized in that: The step S3 further includes: Step S3-1, obtain a vector map of the agricultural product planting area, use a geographic information system to randomly mark points on the vector map, and obtain point coordinates according to a coordinate system; Step S3-2: Acquire soil information for the experimental planting area, including organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, and soil thickness; and obtain meteorological data for the recommended planting area, including rainfall during the growing season, accumulated temperature, average daily minimum temperature, average temperature, and potential evapotranspiration. Step S3-3: randomly select N blocks according to the coordinates of the point, obtain soil information and meteorological data of the N blocks, and form a data set {rainfall during the growing period, accumulated temperature, average daily minimum temperature, average temperature, potential evapotranspiration, soil organic carbon, pH, total nitrogen, total phosphorus, total potassium, cation exchange capacity, soil texture type, soil bulk density, soil thickness}. The elements in the data set are environmental characteristics, and the environmental characteristics are processed using Z-score standardization.

5. The method for agricultural product data recommendation and optimization analysis based on artificial intelligence according to claim 1, characterized in that: The step S4 further includes: Step S4-1: construct a variety recommendation model using multiple linear regression. The variety recommendation model is: in, represents the comprehensive evaluation index of variety adaptability; w is the weight vector, each w j Represents the jth environmental feature x j Forecast results The influence of; b is the bias term, which is a constant term, indicating that when all features x i The output of the model when is zero; the values ​​of w and b are initially set randomly; Step S4-2: Calculate the weight vector w according to the loss function method: Where L(w, b) represents the deviation between the predicted value of w and b and the actual value. The smaller the value, the closer the model prediction is to the actual adaptability. N represents N blocks, W T Represents the rank of the weight vector; y i represents the comprehensive evaluation index of variety adaptability of the i-th phenotype; Step S4-3, calculate the partial derivative of the loss function with respect to the weight: where w j represents the impact of the jth environmental feature on the prediction result, N represents the number of blocks, x ij represents the jth environmental feature of the i-th block, where i is not greater than N; Update the weight w according to the gradient descent method j ; Step S4-4: setting a threshold value for the comprehensive evaluation index of variety adaptability; agricultural products whose comprehensive evaluation index of variety adaptability is greater than the threshold value are considered suitable for promotion.

6. An artificial intelligence-based agricultural product data recommendation optimization analysis system, applied to the artificial intelligence-based agricultural product data recommendation optimization analysis method according to any one of claims 1 to 5, characterized in that: The system includes a phenotypic association analysis module, an adaptability evaluation module, an environmental data acquisition module, and a variety recommendation modeling module; the phenotypic association analysis module is used to analyze the synergistic or antagonistic relationship between agricultural product phenotypes; the adaptability evaluation module is used to calculate the comprehensive evaluation index of agricultural product variety adaptability based on the synergistic or antagonistic relationship; the environmental data acquisition module is used to obtain soil data and meteorological data from the experimental planting area and the recommended planting area of ​​agricultural products to establish a data set; the variety recommendation modeling module is used to construct a variety recommendation model; The output end of the phenotypic association analysis module is electrically connected to the input end of the adaptability evaluation module, the output end of the adaptability evaluation module is electrically connected to the input end of the environmental data acquisition module, and the output end of the environmental data acquisition module is electrically connected to the input end of the variety recommendation modeling module.

7. The artificial intelligence-based agricultural product data recommendation and optimization analysis system according to claim 6, characterized in that: The phenotypic association analysis module includes a phenotypic data acquisition unit and an association relationship analysis unit; the phenotypic data acquisition unit is used to obtain the phenotype of agricultural products from public variety approvals and convert the phenotype into a dimensionless interval value through the mean method; The association relationship analysis unit is used to determine the reference data column according to the dimensionless value, calculate the absolute value of the difference and the association coefficient, and determine the synergistic or antagonistic relationship of the phenotype.

8. The artificial intelligence-based agricultural product data recommendation and optimization analysis system according to claim 6, characterized in that: The adaptability evaluation module includes a data matrix establishment unit and an evaluation index calculation unit; the data matrix establishment unit is used to obtain suitable promotion areas for agricultural products on a provincial basis and establish a data matrix containing the phenotypes of agricultural product varieties in each province; the evaluation index calculation unit is used to standardize the phenotype data and calculate the comprehensive evaluation index of variety adaptability based on the proportion, entropy value and weight of the phenotype.

9. The artificial intelligence-based agricultural product data recommendation and optimization analysis system according to claim 6, characterized in that: The environmental data acquisition module includes a geographic point acquisition unit and an environmental data acquisition unit; the geographic point acquisition unit is used to obtain a vector map of the agricultural product planting area, randomly mark points using a geographic information system and obtain point coordinates; the environmental data acquisition unit is used to obtain soil information and meteorological data of the planting area to form a data set containing environmental characteristics.

10. The artificial intelligence-based agricultural product data recommendation and optimization analysis system according to claim 6, characterized in that: The variety recommendation modeling module includes a model construction unit and a parameter optimization unit; the model construction unit is used to use multivariate linear regression to construct a variety recommendation model, set the initial weights and bias terms of the model, and set the threshold of the comprehensive evaluation index of variety adaptability; the parameter optimization unit is used to calculate the weight vector according to the loss function method and update the weight by the gradient descent method.

Images