Vineyard soil microbial flora regulation and control method and system

By analyzing and preprocessing vineyard soil samples using a fully connected neural network model, a microbial community prediction model was constructed, which solved the problem of the sustainability of soil microbial community regulation and improved the quality and yield of grape growth.

CN119005423APending Publication Date: 2024-11-22NINGXIA UNIVERSITY
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411085089.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Existing technologies make it difficult to continuously monitor and adjust the soil microbial community in vineyards, which affects the quality and yield of grape growth.

Method used

A fully connected neural network model was used to analyze and preprocess vineyard soil samples, and a microbial community prediction model was constructed. The microbial community prediction results were then used for regulation.

Benefits of technology

It enables precise control of the soil microbial environment in vineyards, improves soil structure and fertility, and enhances the health and disease resistance of grapevines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005423A_ABST
    Figure CN119005423A_ABST
Patent Text Reader

Abstract

The invention discloses a vineyard soil microbial flora regulation and control method and system, and belongs to the technical field of microbial flora regulation and control, and the method comprises the following steps: obtaining a vineyard soil sample based on a grape planting range, and analyzing the soil sample to obtain soil data; preprocessing the soil data to obtain a preprocessed data set; constructing a full-connection neural network, improving the full-connection neural network, and training the improved full-connection neural network through a sample set to obtain a flora prediction model; inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result; and regulating and controlling the vineyard soil microbial flora based on the flora prediction result. According to the application, a complex nonlinear relationship can be processed by using the full-connection neural network, so that the model can accurately predict the distribution and possible change of the microbial flora from the soil data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of microbial flora regulation, and in particular relates to a method and system for regulating vineyard soil microbial flora. Background Art

[0002] Microorganisms in the soil play a vital role in the growth of grapes. They can help grape roots absorb nutrients, decompose organic matter, produce hormones needed for growth, and resist certain diseases. By regulating these microorganisms, the growth quality and yield of grapes can be directly affected. Healthy soil microbial communities can improve the structure and fertility of the soil, increase the organic matter content of the soil, and improve the soil's water retention capacity and nutrient cycle efficiency. These are key factors in ensuring the long-term productivity of the soil.

[0003] However, the regulation of soil microorganisms is not a one-time task, but requires continuous monitoring and adjustment, and the response of soil microbial communities may be affected by multiple factors. Therefore, a method and system for regulating vineyard soil microbial flora is urgently needed to solve the above problems. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method and system for regulating vineyard soil microbial flora to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above object, the present invention provides a method for regulating vineyard soil microbial flora, comprising:

[0006] Obtaining soil samples from vineyards based on the grape planting range, analyzing the soil samples, and obtaining soil data;

[0007] Preprocessing the soil data to obtain a preprocessed data set;

[0008] Constructing a fully connected neural network, improving the fully connected neural network and training it with a sample set to obtain a flora prediction model;

[0009] Inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result;

[0010] The vineyard soil microbial flora is regulated based on the flora prediction results.

[0011] Preferably, a process of obtaining a sample of vineyard soil based on a grape planting range, analyzing the sample of soil, and obtaining soil data comprises:

[0012] Divide the vineyard into grids to generate a grid network of the vineyard;

[0013] randomly collecting soil samples through the grid network of the vineyard, and collecting environmental data and intervention records of the grid to which the soil samples belong through sensors;

[0014] Analyze the soil sample by gene sequencing technology to obtain the microbial composition in the sample;

[0015] The environmental data, intervention records and microbial composition in the sample are integrated to obtain the soil data.

[0016] Preferably, the soil data is preprocessed to obtain a preprocessed data set, which includes:

[0017] Performing data cleaning on the soil data, selecting features of the cleaned data, and generating feature selection data;

[0018] The feature selection data is Z-score standardized, and the standardized data is one-hot encoded to generate the preprocessed data set.

[0019] Preferably, the soil data is cleaned, and features of the cleaned data are selected. The process of generating feature selection data includes:

[0020] Deleting missing values ​​and duplicate rows of the soil data, and filtering outliers of the soil data through a box plot to obtain cleaned data;

[0021] The relationship between each feature in the cleaned data and the target variable is evaluated and annotated using the chi-square test method to obtain annotated data;

[0022] The feature weights of the labeled data are evaluated by a random forest method, and data features are selected based on the data weights;

[0023] The data features are exported to obtain the feature selection data.

[0024] Preferably, the process of constructing a fully connected neural network, improving the fully connected neural network and training the neural network with a sample set to obtain a flora prediction model comprises:

[0025] Constructing an initial fully connected neural network, and optimizing the initial fully connected neural network to obtain an optimized model;

[0026] Acquiring a sample set, and amplifying the sample set to obtain an amplified sample set;

[0027] Inputting the amplified sample set into the optimization model for training to obtain a training model;

[0028] Evaluate the performance of the training model on the classification task based on the accuracy index and the precision index to obtain an evaluation result;

[0029] Adjusting the hyperparameters and model structure of the training model based on the evaluation results to obtain an adjusted model;

[0030] The adjustment model is retrained to obtain the flora prediction model.

[0031] Preferably, the process of constructing an initial fully connected neural network and optimizing the initial fully connected neural network to obtain an optimized model includes:

[0032] Constructing the initial fully connected neural network and constructing an attention layer;

[0033] Importing the attention layer into the initial fully connected neural network to generate an attention mechanism model;

[0034] Fine-tune the attention mechanism model on the task of data scheduling balance of microbiome regulation to obtain an improved model;

[0035] The improved model is pre-trained by using the RBMs algorithm, and the training is terminated after the improved model completes the weight distribution learning of the data to generate the optimized model.

[0036] Preferably, the process of obtaining a sample set, amplifying the sample set, and obtaining the amplified sample set includes:

[0037] Building a Generative Adversarial Network;

[0038] Acquire a sample set, and amplify the sample set through the generative adversarial network to obtain an amplified set;

[0039] The amplified set is verified by the Frechet Inception Distance method to generate an amplified sample set.

[0040] Preferably, the amplified sample set is input into the optimization model for training, and the process of obtaining the training model includes:

[0041] Dividing the amplified sample set into a training set and a test set;

[0042] The optimization model is trained using a training set, a cross entropy loss function is selected, and the cross entropy loss function is minimized by a stochastic gradient descent algorithm to obtain a trained model;

[0043] Using the test set to verify the performance of the trained model to obtain a verification result;

[0044] Optimizing the learning rate, number of nodes and number of layers of the trained model based on the verification result to obtain the training model;

[0045] The activation function of the trained model is the Leaky ReLU activation function.

[0046] To achieve the above object, the present invention also provides a vineyard soil microbial flora regulation system, comprising:

[0047] A data acquisition module, used to acquire soil samples of the vineyard based on the grape planting range, analyze the soil samples, and obtain soil data;

[0048] A preprocessing module, used for preprocessing the soil data to obtain a preprocessing data set;

[0049] A network construction module is used to construct a fully connected neural network, and after improving the fully connected neural network, a bacterial flora prediction model is obtained through training with a sample set;

[0050] A prediction module, used for inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result;

[0051] A control module is used to control the vineyard soil microbial flora based on the flora prediction result.

[0052] Compared with the prior art, the present invention has the following advantages and technical effects:

[0053] This application uses a fully connected neural network to handle complex nonlinear relationships, allowing the model to accurately predict the distribution and possible changes of microbial flora from soil data. This high-precision prediction capability is essential for understanding the soil environment and making management decisions. Based on the flora prediction results, the microbial environment in the vineyard soil can be more accurately regulated. For example, by adjusting the nutrients and water conditions of the soil to optimize the growth of specific beneficial flora, the health and disease resistance of the grape plants can be enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0055] Figure 1 This is a flow chart of a method for regulating soil microbial flora in a vineyard according to an embodiment of the present invention;

[0056] Figure 2 Schematic diagram of a vineyard soil microbial flora regulation system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0058] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0059] Embodiment 1

[0060] like Figure 1 As shown, this embodiment provides a method for regulating vineyard soil microbial flora, comprising:

[0061] Obtaining soil samples from vineyards based on the grape planting range, analyzing the soil samples, and obtaining soil data;

[0062] Preprocessing the soil data to obtain a preprocessed data set;

[0063] Constructing a fully connected neural network, improving the fully connected neural network and training it with a sample set to obtain a flora prediction model;

[0064] Inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result;

[0065] The vineyard soil microbial flora is regulated based on the flora prediction results.

[0066] To further optimize the solution, samples of vineyard soil are obtained based on the grape planting area, and the soil samples are analyzed. The process of obtaining soil data includes:

[0067] Divide the vineyard into grids to generate a grid network of the vineyard;

[0068] Before raster division, it is necessary to collect basic geographic data of the vineyard, including land boundaries, topography, soil type, water source distribution, etc. These data can be obtained through geographic information system (GIS) data, satellite images, drone aerial photography, etc.

[0069] The size and shape of the grid are determined according to the management needs and geographical features of the vineyard. The size of the grid can range from a few square meters to hundreds of square meters, and the shape is usually square or rectangular. Soil moisture sensors, temperature sensors, pest and disease monitoring equipment, etc. are installed in the necessary grids to collect data in real time.

[0070] randomly collecting soil samples through the grid network of the vineyard, and collecting environmental data and intervention records of the grid to which the soil samples belong through sensors;

[0071] Analyze the soil sample by gene sequencing technology to obtain the microbial composition in the sample;

[0072] Soil samples need to be physically ground to remove large particles and impurities such as roots to ensure the efficiency and purity of DNA extraction. Use a commercial DNA extraction kit or homemade extraction buffer to extract total microbial DNA from soil samples. The purity and concentration of the DNA are detected by a spectrometer, and the integrity of the DNA is assessed by gel electrophoresis. Depending on the purpose of the study, 16S rRNA gene sequencing (for the identification of bacteria and archaea) or ITS region sequencing (for the identification of fungi) is performed. This includes amplifying specific microbial marker genes, constructing sequencing libraries, and performing high-throughput sequencing. For 16S rRNA gene sequencing, it usually involves the identification of operational taxonomic units (OTUs), species abundance analysis, and functional prediction.

[0073] The environmental data, intervention records and microbial composition in the sample are integrated to obtain the soil data.

[0074] The scheme is further optimized, and the soil data is preprocessed to obtain a preprocessed data set, including:

[0075] Performing data cleaning on the soil data, selecting features of the cleaned data, and generating feature selection data;

[0076] The feature selection data is Z-score standardized, and the standardized data is one-hot encoded to generate the preprocessed data set.

[0077] The expression for Z-score standardization is:

[0078]

[0079] One-hot encoding is a process used to convert categorical features into numerical features. It expands a categorical feature into multiple binary features, where each feature represents a category and takes the value 1 in the category corresponding to the feature and 0 in other cases. For example, suppose a categorical feature has 3 possible values: A,

[0080] B and C, then the one-hot encoded representation is:

[0081] A->[1,0,0];

[0082] B->[0,1,0];

[0083] C->[0,0,1];

[0084] x is the value of the original data point, μ is the mean of the feature, and σ is the standard deviation of the feature.

[0085] To further optimize the scheme, the soil data is cleaned, and features of the cleaned data are selected. The process of generating feature selection data includes:

[0086] Deleting missing values ​​and duplicate rows of the soil data, and filtering outliers of the soil data through a box plot to obtain cleaned data;

[0087] The relationship between each feature in the cleaned data and the target variable is evaluated and annotated using the chi-square test method to obtain annotated data;

[0088] The expression of the chi-square test method is:

[0089]

[0090] Among them, O i is the observed frequency, E i is the expected frequency.

[0091] After calculating the chi-square statistic for all features, you can sort each feature or set a threshold to determine which features are most likely to be related to the target variable.

[0092] The feature weights of the labeled data are evaluated by a random forest method, and data features are selected based on the data weights;

[0093] The labeled data is trained using a random forest model. Random forest assigns a weight to each feature based on its importance.

[0094] Commonly used feature importance evaluation indicators in random forests include Gini Importance and Mean Decrease Impurity.

[0095] Gini importance is usually calculated using the weighted average of the Gini index of each feature's split points in all decision trees.

[0096] According to the calculated feature weights, features with higher weights can be selected as the final selected data features.

[0097] The data features are exported to obtain the feature selection data.

[0098] Further optimizing the scheme, the process of constructing a fully connected neural network, improving the fully connected neural network and obtaining a flora prediction model through sample set training includes:

[0099] Constructing an initial fully connected neural network, and optimizing the initial fully connected neural network to obtain an optimized model;

[0100] Acquiring a sample set, and amplifying the sample set to obtain an amplified sample set;

[0101] Inputting the amplified sample set into the optimization model for training to obtain a training model;

[0102] Evaluate the performance of the training model on the classification task based on the accuracy index and the precision index to obtain an evaluation result;

[0103] Adjusting the hyperparameters and model structure of the training model based on the evaluation results to obtain an adjusted model;

[0104] The adjustment model is retrained to obtain the flora prediction model.

[0105] Further optimizing the scheme, the process of constructing an initial fully connected neural network and optimizing the initial fully connected neural network to obtain an optimized model includes:

[0106] Constructing the initial fully connected neural network and constructing an attention layer;

[0107] Import the attention layer into the initial fully connected neural network to generate an attention mechanism model; insert the attention mechanism in the middle or last layer of the model so that the model can better understand and utilize the relationship between the input data. When the input data is forward propagated through the fully connected neural network, when it reaches the inserted fully connected layer, part of the data will enter the attention mechanism, and the other part of the data will continue to be passed to the next fully connected layer. The attention mechanism layer usually has its own parameters, such as the weight matrix of Query, Key, Value, etc. These parameters will be trained together with other parameters of the fully connected neural network.

[0108] Fine-tune the attention mechanism model on the task of data scheduling balance of microbiome regulation to obtain an improved model;

[0109] The improved model is pre-trained by using the RBMs algorithm, and the training is terminated after the improved model completes the weight distribution learning of the data to generate the optimized model.

[0110] The RBMs algorithm can be expressed as:

[0111]

[0112] in, is the state vector of the visible layer, is the state vector of the hidden layer, W ij is the connected visible unit v i and hidden units h j The weight of b i is the visible unit v i The bias term, c j is the hidden unit h j The bias term.

[0113] Further optimizing the scheme, obtaining a sample set, amplifying the sample set, and obtaining the amplified sample set includes the following steps:

[0114] Building a Generative Adversarial Network;

[0115] Acquire a sample set, and amplify the sample set through the generative adversarial network to obtain an amplified set;

[0116] The amplified set is verified by the Frechet Inception Distance method to generate an amplified sample set.

[0117] Further optimizing the scheme, inputting the amplified sample set into the optimization model for training, the process of obtaining the training model includes:

[0118] Dividing the amplified sample set into a training set and a test set;

[0119] The optimization model is trained using a training set, a cross entropy loss function is selected, and the cross entropy loss function is minimized by a stochastic gradient descent algorithm to obtain a trained model;

[0120] Using the test set to verify the performance of the trained model to obtain a verification result;

[0121] Optimizing the learning rate, number of nodes and number of layers of the trained model based on the verification result to obtain the training model;

[0122] The activation function of the trained model is the Leaky ReLU activation function.

[0123] The expression of the Leaky ReLU activation function is:

[0124] f(x) = max(αx, x);

[0125] Here x is the input and α is a small positive constant (such as 0.01) used to provide a non-zero slope for the negative input portion.

[0126] Embodiment 2

[0127] like Figure 2As shown, this embodiment provides a vineyard soil microbial flora regulation system, comprising:

[0128] A data acquisition module, used to acquire soil samples of the vineyard based on the grape planting range, analyze the soil samples, and obtain soil data;

[0129] A preprocessing module, used for preprocessing the soil data to obtain a preprocessing data set;

[0130] A network construction module is used to construct a fully connected neural network, and after improving the fully connected neural network, a bacterial flora prediction model is obtained through training with a sample set;

[0131] A prediction module, used for inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result;

[0132] A control module is used to control the vineyard soil microbial flora based on the flora prediction result.

[0133] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for regulating vineyard soil microbial flora, characterized in that: The following steps are involved: Obtaining soil samples from vineyards based on the grape planting range, analyzing the soil samples, and obtaining soil data; Preprocessing the soil data to obtain a preprocessed data set; Constructing a fully connected neural network, improving the fully connected neural network and training it with a sample set to obtain a flora prediction model; Inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result; The vineyard soil microbial flora is regulated based on the flora prediction results.

2. The method for regulating vineyard soil microbial flora according to claim 1, characterized in that: The process of obtaining soil samples of vineyards based on the grape planting area and analyzing the soil samples to obtain soil data includes: Divide the vineyard into grids to generate a grid network of the vineyard; randomly collecting soil samples through the grid network of the vineyard, and collecting environmental data and intervention records of the grid to which the soil samples belong through sensors; Analyze the soil sample by gene sequencing technology to obtain the microbial composition in the sample; The environmental data, intervention records and microbial composition in the sample are integrated to obtain the soil data.

3. The method for regulating vineyard soil microbial flora according to claim 1, characterized in that: The process of preprocessing the soil data to obtain a preprocessed data set includes: Performing data cleaning on the soil data, selecting features of the cleaned data, and generating feature selection data; The feature selection data is Z-score standardized, and the standardized data is one-hot encoded to generate the preprocessed data set.

4. The method for regulating vineyard soil microbial flora according to claim 3, characterized in that: The soil data is cleaned, and features of the cleaned data are selected. The process of generating feature selection data includes: Deleting missing values ​​and duplicate rows of the soil data, and filtering outliers of the soil data through a box plot to obtain cleaned data; The relationship between each feature in the cleaned data and the target variable is evaluated and annotated using the chi-square test method to obtain annotated data; The feature weights of the labeled data are evaluated by a random forest method, and data features are selected based on the data weights; The data features are exported to obtain the feature selection data.

5. The method for regulating vineyard soil microbial flora according to claim 1, characterized in that: The process of constructing a fully connected neural network, improving the fully connected neural network and training the neural network with a sample set to obtain a flora prediction model includes: Constructing an initial fully connected neural network, and optimizing the initial fully connected neural network to obtain an optimized model; Acquiring a sample set, and amplifying the sample set to obtain an amplified sample set; Inputting the amplified sample set into the optimization model for training to obtain a training model; Evaluate the performance of the training model on the classification task based on the accuracy index and the precision index to obtain an evaluation result; Adjusting the hyperparameters and model structure of the training model based on the evaluation results to obtain an adjusted model; The adjustment model is retrained to obtain the flora prediction model.

6. The method for regulating vineyard soil microbial flora according to claim 5, characterized in that: The process of constructing an initial fully connected neural network and optimizing the initial fully connected neural network to obtain an optimized model includes: Constructing the initial fully connected neural network and constructing an attention layer; Importing the attention layer into the initial fully connected neural network to generate an attention mechanism model; Fine-tune the attention mechanism model on the task of data scheduling balance of microbiome regulation to obtain an improved model; The improved model is pre-trained by using the RBMs algorithm, and the training is terminated after the improved model completes the weight distribution learning of the data to generate the optimized model.

7. The method for regulating vineyard soil microbial flora according to claim 5, characterized in that: The process of obtaining a sample set and amplifying the sample set to obtain the amplified sample set includes: Building a Generative Adversarial Network; Acquire a sample set, and amplify the sample set through the generative adversarial network to obtain an amplified set; The amplified set is verified by the Frechet Inception Distance method to generate an amplified sample set.

8. The method for regulating vineyard soil microbial flora according to claim 5, characterized in that: The amplified sample set is input into the optimization model for training. The process of obtaining the training model includes: Dividing the amplified sample set into a training set and a test set; The optimization model is trained using a training set, a cross entropy loss function is selected, and the cross entropy loss function is minimized by a stochastic gradient descent algorithm to obtain a trained model; Using the test set to verify the performance of the trained model to obtain a verification result; Optimizing the learning rate, number of nodes and number of layers of the trained model based on the verification result to obtain the training model; The activation function of the trained model is the Leaky ReLU activation function.

9. A vineyard soil microbial flora control system, characterized in that: include: A data acquisition module, used to acquire soil samples of the vineyard based on the grape planting range, analyze the soil samples, and obtain soil data; A preprocessing module, used for preprocessing the soil data to obtain a preprocessing data set; A network construction module is used to construct a fully connected neural network, and after improving the fully connected neural network, a bacterial flora prediction model is obtained through training with a sample set; A prediction module, used for inputting the preprocessed data set into the flora prediction model for calculation to obtain a flora prediction result; A control module is used to control the vineyard soil microbial flora based on the flora prediction result.

Citation Information

Cited By

  • Soil micro-ecology flora regulation and control method and system

    CN120832827A

  • Method and system for regulating the microflora of the soil

    CN120832827B