High-standard farmland full-process intelligent information management algorithm system and method

Abstract

The invention relates to the technical field of agricultural information, in particular to a high-standard farmland full-process intelligent information management algorithm system and method. The invention provides a high-standard farmland full-process intelligent information management algorithm system, which comprises a data acquisition module, a data processing module, a feature extraction module, a pest and disease damage and mildew early warning model construction module and an early warning information generation and push module, and is characterized in that the data acquisition module is used for acquiring temperature and humidity data and image data; the data processing module is used for data cleaning, the feature extraction module is used for extracting change features and appearance features of temperature and humidity data, and the disease and pest and mildew early warning model construction module adopts a machine learning algorithm to construct an early warning model. According to the method, through multi-dimensional data feature extraction, environmental data are considered, color, texture and surface anomaly features are extracted from the agricultural product appearance image data, the state of the agricultural product can be evaluated more comprehensively, and the early warning accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural information technology, and in particular to a full-process intelligent information management algorithm system and method for high-standard farmland. Background Art

[0002] In modern agricultural production, the construction of high-standard farmland has become a key link in ensuring food security and the quality of agricultural products, and agricultural product storage management is a vital part of the entire agricultural industry chain.

[0003] During the storage of agricultural products, pests and diseases and mildew problems seriously affect the quality and storage safety of agricultural products. Traditional monitoring methods often rely on regular manual inspections, which is not only inefficient, but also difficult to capture the early signs of pests and diseases and mildew in real time and comprehensively. Since potential risks cannot be discovered in a timely and accurate manner, when pests and diseases or mildew break out on a large scale, it will cause huge losses to the storage of agricultural products. Therefore, there is an urgent need for an intelligent system that can monitor pests and diseases and mildew in real time and accurately warn of them. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a full-process intelligent information management algorithm system and method for high-standard farmland, which can effectively solve the problems mentioned in the prior art.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0006] The present invention provides a full-process intelligent information management algorithm system for high-standard farmland, which is characterized by comprising a data acquisition module, a data processing module, a feature extraction module, a pest and mildew early warning model construction module and an early warning information generation and push module; the data acquisition module comprises a sensor unit and a monitoring unit, the sensor unit is used to collect temperature and humidity data of agricultural product storage environment, and the monitoring unit is used to collect agricultural product appearance image data; the data processing module is used to clean the collected temperature and humidity data, remove abnormal values ​​and erroneous data, and perform denoising on the image data to improve image quality; the feature extraction module is used to extract the change characteristics of temperature and humidity data, as well as the appearance characteristics of agricultural products; the pest and mildew early warning model construction module adopts a machine learning algorithm to construct an early warning model, uses historical data to train the model, and has an automatic update mechanism to adapt to environmental changes; the early warning information generation and push module generates early warning information including pest or mildew type, expected occurrence time and treatment suggestions according to the judgment result of the early warning model, and pushes it to warehouse managers through multiple communication methods.

[0007] Furthermore, the sensor unit includes a plurality of temperature sensors and humidity sensors; and the monitoring unit includes a plurality of high-definition cameras.

[0008] Furthermore, the data processing module removes abnormal values ​​of temperature and humidity data based on the 3σ principle; assuming that the temperature data sequence is T={t 1 , t 2 , ..., t 3}, calculate the mean The formula is:

[0009]

[0010] Calculate the standard deviation using the formula:

[0011]

[0012] According to the 3σ principle, or The temperature data is considered as an outlier and removed; let the humidity data sequence be H = {h 1 ,h 2 , ..., h 3}, calculate the mean The formula is:

[0013]

[0014] Calculate the standard deviation using the formula:

[0015]

[0016] According to the 3σ principle, or The humidity data are considered as outliers and removed.

[0017] The data processing module uses mean filtering to perform denoising on the image data; suppose the image I(x, y), the filter window size is (2k+1)×(2k+1), and the image I after mean filtering f The calculation formula for (x, y) is:

[0018]

[0019] Furthermore, the feature extraction module extracts the change features of the temperature and humidity data by combining variational mode decomposition with the slope change rate; the temperature and humidity data are set as the environmental data X = {x 1 , x 2 , ..., x n}, perform variational mode decomposition on X, and decompose X into k modes u by solving the following constrained variational problem k (t) and the corresponding center frequency w k , the calculation formula is:

[0020]

[0021] in, is the partial derivative of t, δ(t) is the Dirac function, and * is the convolution; the solved mode u k (t) satisfies the following formula:

[0022]

[0023] Calculate the slope change rate of each mode, for each mode u k , use the least squares method to fit the straight line u k (t), the calculation formula is:

[0024] u k (t) = a k (t)+b k ;

[0025] Among them, a k is the changing trend of the mode; then calculate the slope change rate Δa k , assuming the time interval is Δt, the calculation formula is:

[0026]

[0027] Then, based on the modal energy, the fluctuation amplitude characteristics are obtained and the energy E of each mode is calculated. k , the calculation formula is:

[0028]

[0029] The fluctuation amplitude characteristics are expressed by the standard deviation of each mode energy, and the calculation formula is:

[0030]

[0031] Furthermore, the feature extraction module uses a multi-scale fusion convolutional neural network to extract appearance features of agricultural products, and the appearance features include color features, texture features and surface abnormality features.

[0032] Color feature extraction, assuming that the input image is I(x, y), the size is H×W×C, H is the height, W is the width, C is the number of channels, and the convolution kernel of the cth channel in the first layer is i and j represent the position of the convolution kernel, and the size is k h ×k w , perform convolution operation on the input image to obtain the c-th channel feature map of the l+1th layer The calculation formula is:

[0033]

[0034] in, is the pixel value of the m-th channel image at (x, y) in the l-th layer, is the bias; perform global average pooling, and assume that the feature map obtained after L layers of convolution is Then the color feature vector The calculation formula is:

[0035]

[0036] Where H is the image height and W is the image width.

[0037] Texture feature extraction, suppose after n layers of convolution and pooling, we get a feature map T(x, y), the size is h×w×d, d is the depth of the feature map, the number of gray levels is G, and the distance parameter is d 0 , angle θ, calculate the gray level co-occurrence matrix P ij (d 0 ,θ), the calculation formula is:

[0038]

[0039] Then, the texture features are quantified by contrast. The formula for calculating contrast is:

[0040]

[0041] Surface abnormality feature extraction, assuming the denoised image is The original image is I(x, y), and the difference convolution kernel is The size is k hD ×K wD , perform difference convolution operation to obtain the difference feature map F D (x, y), the calculation formula is:

[0042]

[0043] Among them, b D is the bias; after N layers of difference convolution and pooling, the feature map F is obtained DN (x, y), mapped to the surface anomaly feature vector V through the fully connected layer A , the calculation formula is:

[0044]

[0045] Among them, W A is the weight matrix, b A is the bias, h N , W N and D N They are the height, width and depth of the feature map after the last layer of difference convolution and pooling, respectively.

[0046] Furthermore, the pest and mildew early warning model construction module adopts a hybrid neural network based on the attention mechanism, which includes multiple branches, which are used to process temperature and humidity data characteristics and agricultural product appearance characteristics respectively.

[0047] Dynamic weighted data fusion is performed. The change characteristics of temperature and humidity data and the importance of appearance characteristics to pest and mold early warning change with the changes in the environment and the state of agricultural products. A dynamic weighting mechanism is introduced. The characteristic vector of temperature and humidity change is X TH , the appearance feature is X A , calculate the dynamic weight w at the input layer of the hybrid neural network TH and w A , the fused input vector X f The calculation formula is:

[0048] Xi=w TH X TH +w A X A ;

[0049] Then the model training is carried out, and the fused feature vector X f Input into the constructed neural network, after calculation of multiple hidden layers, the prediction result y is obtained p , including the probability of occurrence of pests and diseases and the degree of mildew; the cross entropy loss function is used to measure the difference between the predicted results and the true labels. The formula of the cross entropy loss function is:

[0050]

[0051] Among them, y p is the true label; use the Adam optimizer to update the network weights and bias parameters according to the loss function, and calculate the gradient g of the loss function to each parameter. The formula is:

[0052]

[0053] Among them, θ is the network parameter, which is the update parameter of the Adam optimizer.

[0054] The pest and mildew early warning model is updated by using an incremental learning method, which combines new historical data, including new temperature and humidity data change characteristics and agricultural product appearance characteristics, with the original training data into a new training set and retrains the model.

[0055] The full-process intelligent information management method of high-standard farmland is applied to the full-process intelligent information management algorithm system of high-standard farmland, and includes the following steps:

[0056] Step 1: Collect agricultural product storage environment data and agricultural product appearance data through the data collection module; Step 2: Use the data processing and feature extraction module to clean, calibrate and extract features from the collected data; Step 3: Use the warning model constructed by the pest and mildew warning model construction module to analyze the feature data and determine the possibility of pests and mildew; Step 4: Generate and push warning information through the warning information generation and push module, and the management personnel take corresponding measures according to the processing suggestions and update the warning model regularly.

[0057] Furthermore, the warning information includes specific types of pests or mildew.

[0058] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects:

[0059] The present invention can collect data on the storage environment and appearance of agricultural products in real time through the sensor unit and the appearance monitoring unit. Compared with the traditional manual inspection method, the monitoring frequency is greatly improved. The sensor collects environmental data and the camera takes images of the appearance of agricultural products. It can capture environmental changes and subtle changes in the appearance of agricultural products in time, providing a data basis for early warning of pests and diseases and mildew.

[0060] The present invention not only considers environmental data, such as temperature change rate and humidity fluctuation amplitude, through multi-dimensional data feature extraction, but also extracts color, texture and surface abnormality features from agricultural product appearance image data. This multi-dimensional data comprehensive analysis can more comprehensively evaluate the status of agricultural products and improve the accuracy of early warning.

[0061] The present invention constructs an early warning model, trains it through a large amount of historical data, and can regularly update the model based on new data. These models can learn complex data patterns and rules, and accurately predict the types and occurrence times of pests and mildew, effectively avoiding omissions or false alarms caused by human judgment errors in traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0065] The present invention will be further described below in conjunction with the embodiments.

[0066] Example:

[0067] The full-process intelligent information management algorithm system for high-standard farmland is characterized by including a data acquisition module, a data processing module, a feature extraction module, a pest and mildew early warning model construction module and an early warning information generation and push module.

[0068] The data acquisition module includes a sensor unit and a monitoring unit. The sensor unit is used to collect temperature and humidity data of the agricultural product storage environment, and the monitoring unit is used to collect image data of the appearance of agricultural products. The data processing module is used to clean the collected temperature and humidity data, remove outliers and erroneous data, and denoise the image data to improve image quality. The feature extraction module is used to extract the changing characteristics of temperature and humidity data, as well as the appearance characteristics of agricultural products. The pest and mildew early warning model construction module uses a machine learning algorithm to build an early warning model, uses historical data to train the model, and has an automatic update mechanism to adapt to environmental changes. The early warning information generation and push module generates early warning information including pest or mildew type, expected occurrence time and treatment suggestions according to the judgment results of the early warning model, and pushes it to warehouse managers through various communication methods.

[0069] Furthermore, the sensor unit includes a plurality of temperature sensors and humidity sensors; and the monitoring unit includes a plurality of high-definition cameras.

[0070] Furthermore, the data processing module removes outliers in temperature and humidity data based on the 3σ principle;

[0071] Assume that the temperature data sequence is T = {t 1 , t 2 , ..., t 3}, calculate the mean The formula is:

[0072]

[0073] Calculate the standard deviation using the formula:

[0074]

[0075] According to the 3σ principle, or The temperature data is considered as an outlier and removed; let the humidity data sequence be H = {h 1 ,h 2 , ..., h 3}, calculate the mean The formula is:

[0076]

[0077] Calculate the standard deviation using the formula:

[0078]

[0079] According to the 3σ principle, or The humidity data are considered as outliers and removed.

[0080] The data processing module uses mean filtering to denoise the image data; suppose the image I(x, y), the filter window size is (2k+1)×(2k+1), and the image I after mean filtering is f The calculation formula for (x, y) is:

[0081]

[0082] Furthermore, the feature extraction module extracts the change characteristics of the temperature and humidity data by combining variational mode decomposition with the slope change rate; the temperature and humidity data are set as the environmental data X = {x 1 , x 2 , ..., x n}, perform variational mode decomposition on X, and decompose X into k modes u by solving the following constrained variational problem k (t) and the corresponding center frequency w k , the calculation formula is:

[0083]

[0084] in, is the partial derivative of t, δ(t) is the Dirac function, and * is the convolution;

[0085] The solved mode u k (t) satisfies the following formula:

[0086]

[0087] Calculate the slope change rate of each mode, for each mode u k , use the least squares method to fit the straight line u k (t), the calculation formula is:

[0088] u k (t) = a k (t)+b k ;

[0089] Among them, a k is the changing trend of the mode; then calculate the slope change rate Δa k , assuming the time interval is Δt, the calculation formula is:

[0090]

[0091] Then, based on the modal energy, the fluctuation amplitude characteristics are obtained and the energy E of each mode is calculated. k , the calculation formula is:

[0092]

[0093] The fluctuation amplitude characteristics are expressed by the standard deviation of each mode energy, and the calculation formula is:

[0094]

[0095] Furthermore, the feature extraction module uses a multi-scale fusion convolutional neural network to extract the appearance features of agricultural products, which include color features, texture features and surface abnormality features.

[0096] Color feature extraction, assuming that the input image is I(x, y), the size is H×W×C, H is the height, W is the width, C is the number of channels, and the convolution kernel of the cth channel of the lth layer is i and j represent the position of the convolution kernel, and the size is k h ×k w , perform convolution operation on the input image to obtain the c-th channel feature map of the l+1th layer The calculation formula is:

[0097]

[0098] in, is the pixel value of the m-th channel image at (x, y) in the l-th layer, is the bias; perform global average pooling, and assume that the feature map obtained after L layers of convolution is Then the color feature vector The calculation formula is:

[0099]

[0100] Among them, H is the image height and W is the image width.

[0101] Texture feature extraction, suppose after n layers of convolution and pooling, we get a feature map T(x, y), the size is h×w×d, d is the depth of the feature map, the number of gray levels is G, and the distance parameter is d 0 , angle θ, calculate the gray level co-occurrence matrix P ij (d 0 ,θ), the calculation formula is:

[0102]

[0103] Then, the texture features are quantified by contrast. The formula for calculating contrast is:

[0104]

[0105] Surface abnormality feature extraction, assuming the denoised image is The original image is I(x, y), and the difference convolution kernel is The size is k hD ×k wD , perform difference convolution operation to obtain the difference feature map F D (x, y), the calculation formula is:

[0106]

[0107] Among them, b D is the bias; after N layers of difference convolution and pooling, the feature map F is obtained DN (x, y), mapped to the surface anomaly feature vector V through the fully connected layer A , the calculation formula is:

[0108]

[0109] Among them, W A is the weight matrix, b A is the bias, h N 、w N and D N They are the height, width and depth of the feature map after the last layer of difference convolution and pooling, respectively.

[0110] Furthermore, the pest and mildew early warning model construction module adopts a hybrid neural network based on the attention mechanism, which contains multiple branches, which are used to process temperature and humidity data characteristics and agricultural product appearance characteristics respectively.

[0111] Dynamic weighted data fusion is performed. The change characteristics of temperature and humidity data and the importance of appearance characteristics to pest and mold early warning change with the changes in the environment and the state of agricultural products. A dynamic weighting mechanism is introduced. The characteristic vector of temperature and humidity change is X TH , the appearance feature is X A, calculate the dynamic weight w at the input layer of the hybrid neural network TH and w A , the fused input vector X f The calculation formula is:

[0112] X f =w TH X TH +w A X A ;

[0113] Then the model training is carried out, and the fused feature vector X f Input into the constructed neural network, after calculation of multiple hidden layers, the prediction result y is obtained p , including the probability of occurrence of pests and diseases and the degree of mildew; the cross entropy loss function is used to measure the difference between the predicted results and the true labels. The formula of the cross entropy loss function is:

[0114]

[0115] Among them, y p is the true label; use the Adam optimizer to update the network weights and bias parameters according to the loss function, and calculate the gradient g of the loss function to each parameter. The formula is:

[0116]

[0117] Among them, θ is the network parameter, which is the update parameter of the Adam optimizer.

[0118] The pest and mildew early warning model is updated using an incremental learning method, which combines new historical data, including new temperature and humidity data change characteristics and agricultural product appearance characteristics, with the original training data into a new training set and retrains the model.

[0119] The full-process intelligent information management method of high-standard farmland is applied to the full-process intelligent information management algorithm system of high-standard farmland, and includes the following steps: Step 1, collecting agricultural product storage environment data and agricultural product appearance data through the data acquisition module; Step 2, using the data processing and feature extraction module to clean, calibrate and extract features from the collected data; Step 3, using the early warning model constructed by the pest and mildew early warning model construction module to analyze the feature data to determine the possibility of the occurrence of pests and diseases or mildew; Step 4, generating and pushing early warning information through the early warning information generation and push module, and the management personnel take corresponding measures according to the processing suggestions and regularly update the early warning model.

[0120] Furthermore, the warning information includes specific types of pests, diseases or mildew.

[0121] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. The whole process intelligent information management algorithm system of high-standard farmland is characterized by: It includes data collection module, data processing module, feature extraction module, pest and mildew early warning model construction module and early warning information generation and push module; The data acquisition module includes a sensor unit and a monitoring unit, wherein the sensor unit is used to collect temperature and humidity data of the agricultural product storage environment, and the monitoring unit is used to collect image data of the agricultural product appearance; The data processing module is used to clean the collected temperature and humidity data, remove abnormal values ​​and erroneous data, and perform denoising on the image data to improve the image quality; The feature extraction module is used to extract the change characteristics of temperature and humidity data, as well as the appearance characteristics of agricultural products; The pest and mildew early warning model construction module uses a machine learning algorithm to build an early warning model, uses historical data to train the model, and has an automatic update mechanism to adapt to environmental changes; The warning information generation and push module generates warning information including the type of pests or mildew, expected occurrence time and treatment suggestions according to the judgment results of the warning model, and pushes it to the warehouse management personnel through various communication methods.

2. The high-standard farmland full-process intelligent information management algorithm system according to claim 1 is characterized in that: The sensor unit includes a plurality of temperature sensors and humidity sensors; the monitoring unit includes a plurality of high-definition cameras.

3. The high-standard farmland full-process intelligent information management algorithm system according to claim 1 is characterized in that: The data processing module removes abnormal values ​​of temperature and humidity data based on the 3σ principle; Assume that the temperature data sequence is T = {t1, t2, ..., t3}, calculate the mean The formula is: Calculate the standard deviation using the formula: According to the 3σ principle, or The temperature data are considered as outliers and removed; Assume that the humidity data sequence is H = {h1, h2, ..., h3}, calculate the mean The formula is: Calculate the standard deviation using the formula: According to the 3σ principle, or The humidity data are considered as outliers and removed; The data processing module uses mean filtering to perform denoising on the image data; Suppose image I(x, y), filter window size is (2k+1)×(2k+1), image I after mean filtering f The calculation formula for (x, y) is:

4. The high-standard farmland full-process intelligent information management algorithm system according to claim 1 is characterized in that: The feature extraction module extracts the change features of temperature and humidity data by combining variational mode decomposition with slope change rate; The temperature and humidity data are set as the environmental data X = {x1, x2, ..., x n }, perform variational mode decomposition on X, and decompose X into k modes u by solving the following constrained variational problem k (t) and the corresponding center frequency w k , the calculation formula is: in, is the partial derivative of t, δ(t) is the Dirac function, and * is the convolution; The solved mode u k (t) satisfies the following formula: Calculate the slope change rate of each mode, for each mode u k , use the least squares method to fit the straight line u k (t), the calculation formula is: u k (t)=a k (t)+b k ; Among them, a k is the changing trend of the mode; Then calculate the slope change rate Δa k , assuming the time interval is Δt, the calculation formula is: Then, based on the modal energy, the fluctuation amplitude characteristics are obtained and the energy E of each mode is calculated. k , the calculation formula is: The fluctuation amplitude characteristics are expressed by the standard deviation of each mode energy, and the calculation formula is:

5. The high-standard farmland full-process intelligent information management algorithm system according to claim 1 is characterized in that: The feature extraction module uses a multi-scale fusion convolutional neural network to extract the appearance features of agricultural products, and the appearance features include color features, texture features and surface abnormality features; Color feature extraction, assuming that the input image is I(x, y), the size is H×W×C, H is the height, W is the width, C is the number of channels, and the convolution kernel of the cth channel of the lth layer is i and j represent the position of the convolution kernel, and the size is k h ×k w , perform convolution operation on the input image to obtain the c-th channel feature map of the l+1th layer The calculation formula is: in, is the pixel value of the m-th channel image at (x, y) in the l-th layer, is bias; Perform global average pooling, and assume that the feature map obtained after L layers of convolution is Then the color feature vector The calculation formula is: Where H is the image height and W is the image width; Texture feature extraction, suppose after n layers of convolution and pooling, the feature map T(x, y) is obtained, the size is h×w×d, d is the depth of the feature map, the gray level is G, the distance parameter is d0, the angle is θ, and the gray level co-occurrence matrix P is calculated. ij (d0, θ), the calculation formula is: Then, the texture features are quantified by contrast. The formula for calculating contrast is: Surface abnormality feature extraction, assuming the denoised image is The original image is I(x, y), and the difference convolution kernel is The size is k hD ×k wD , perform difference convolution operation to obtain the difference feature map F D (x, y), the calculation formula is: Among them, b D is bias; After N layers of difference convolution and pooling, the feature map F is obtained. DN (x, y), mapped to the surface anomaly feature vector V through the fully connected layer A , the calculation formula is: Among them, W A is the weight matrix, b A is the bias, h N 、w N and D N They are the height, width and depth of the feature map after the last layer of difference convolution and pooling, respectively.

6. The high-standard farmland full-process intelligent information management algorithm system according to claim 1 is characterized in that: The pest and mildew early warning model construction module adopts a hybrid neural network based on the attention mechanism, which includes multiple branches, which are used to process temperature and humidity data features and agricultural product appearance features respectively; Dynamic weighted data fusion is performed. The change characteristics of temperature and humidity data and the importance of appearance characteristics to pest and mold early warning change with the changes in the environment and the state of agricultural products. A dynamic weighting mechanism is introduced. The characteristic vector of temperature and humidity change is X TH , the appearance feature is X A , calculate the dynamic weight w at the input layer of the hybrid neural network TH and w A , the fused input vector X f The calculation formula is: X f =w TH X TH +w A X A ; Then the model training is carried out, and the fused feature vector X f Input into the constructed neural network, after calculation of multiple hidden layers, the prediction result y is obtained p , including the probability of pests and diseases and the degree of mildew; The cross entropy loss function is used to measure the difference between the predicted result and the true label. The formula of the cross entropy loss function is: Among them, y p is the true label; Use the Adam optimizer to update the network weights and bias parameters according to the loss function, and calculate the gradient g of the loss function to each parameter. The formula is: Among them, θ is the network parameter, which is the update parameter of the Adam optimizer; The pest and mildew early warning model is updated by using an incremental learning method, which combines new historical data, including new temperature and humidity data change characteristics and agricultural product appearance characteristics, with the original training data into a new training set and retrains the model.

7. A method for intelligent information management of the entire process of high-standard farmland, applied to the algorithm system for intelligent information management of the entire process of high-standard farmland according to any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: Collect agricultural product storage environment data and agricultural product appearance data through a data collection module; Step 2: Use the data processing and feature extraction module to clean, calibrate and extract features from the collected data; Step 3: Use the early warning model constructed by the pest and mildew early warning model construction module to analyze the characteristic data and determine the possibility of pest and mildew occurrence; Step 4: Generate and push warning information through the warning information generation and push module. Managers take corresponding measures based on processing suggestions and update the warning model regularly.

8. The high-standard farmland full-process intelligent information management method according to claim 7 is characterized in that: The warning information includes specific pests, diseases or mildew types.

Images