A smart agricultural management method and system applied to ecological agriculture

Abstract

The present invention discloses a smart agricultural management method and system applied to ecological agriculture, and relates to the field of ecological agricultural management technology. The system can optimize model parameters and feature selection by continuously feeding back prediction results and actual situations. This closed-loop mechanism ensures the continuous improvement of the system and further improves the accuracy of prediction and decision-making. By refining feature extraction and selection, the sensitivity of the model to key influencing factors is improved, and the reliability of the prediction results is enhanced. By utilizing the integrated learning ability of random forests, the model remains efficient in complex environments and reduces dependence on a single feature or model. The comprehensive scoring system makes pesticide application and management decisions more data-driven and reduces the waste of resources caused by empirical judgment. The system's dynamic adjustment mechanism reduces unnecessary chemical inputs, helps protect the ecological environment, and promotes the development of ecological agriculture.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological agriculture management, and in particular to a smart agriculture management method and system applied to ecological agriculture. Background Art

[0002] In the field of modern agriculture, intelligence and digitalization have become a development trend. Ecological agriculture, as an important branch, aims to achieve sustainable agricultural production. The introduction of smart agricultural management systems not only improves production efficiency, but also optimizes resource allocation, and promotes the development of agriculture in a more environmentally friendly and efficient direction. In this context, prediction models based on meteorological and biological data have emerged as the times require, becoming an important tool for identifying potential pest and disease risks, and further promoting the sustainable development of ecological agriculture.

[0003] Although the existing ecological agricultural management system has made some progress in pest and disease monitoring and management, it still faces many challenges. For example, traditional pest and disease identification methods often rely on experience and lack scientific basis, resulting in the inability to timely and accurately predict the occurrence of pests and diseases.

[0004] These deficiencies are mainly due to the lag in data processing and analysis, which leads to a lack of effective decision-making support for farmers in the face of complex climatic conditions and the variability of pests and diseases. As a result, not only the healthy growth of crops is affected, but also excessive or insufficient use of pesticides may affect the ecological safety of soil and water bodies. This vicious cycle not only affects the yield and quality of crops, but may also have long-term negative impacts on the ecosystem. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a smart agriculture management method and system applied to ecological agriculture, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a smart agricultural management system applied to ecological agriculture, including a data acquisition module, a data preprocessing module, a risk assessment module, a prediction model module, a decision support module and a feedback and adjustment module;

[0007] The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology;

[0008] The data preprocessing module performs data cleaning and normalization processing on the meteorological data M, biological data B and insect pest data PI collected by the data collection module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PIS ;

[0009] The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R;

[0010] The prediction model module uses a random forest machine learning algorithm to perform risk prediction on the pest risk value R and predict the probability of pest occurrence P;

[0011] The decision support module provides pesticide application and management suggestions based on the probability of pest occurrence P, and comprehensively considers ecological agriculture by using the comprehensive score S;

[0012] The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

[0013] Preferably, the data collection module includes a meteorological data collection unit and a biological data and pest data collection unit;

[0014] The meteorological data acquisition unit collects meteorological data M including air temperature T, air humidity H and rainfall O through sensors, wherein the sensors include temperature sensors, humidity sensors and rain gauges;

[0015] The biological data and pest data acquisition unit collects biological data B and pest data PI through a plant growth monitor, a high-definition camera, an insect trap and image recognition technology, wherein the biological data B includes plant growth conditions A and soil moisture F, and the plant growth conditions A includes plant height and stem width.

[0016] Preferably, the data preprocessing module includes a data cleaning unit and a data normalization unit;

[0017] The data cleaning unit cleans the meteorological data M, biological data B and pest data PI by detecting outliers and filling missing values, eliminating outliers and missing values, and obtaining the cleaned meteorological data M. A 、Biological Data B A and Pest Data PI A ;

[0018] The data normalization unit performs the cleaned meteorological data M A 、Biological Data B A and Pest Data PI A Perform normalization processing to obtain the processed meteorological data M S , processed biological data B SAnd the treated pest data PI S , making it convenient for subsequent analysis under the same dimension.

[0019] Preferably, the risk assessment module includes a data input unit and a risk assessment calculation unit;

[0020] The data input unit receives the meteorological data M obtained and processed from the data preprocessing module. S , processed biological data B S And the treated pest data PI S , input into the linear regression prediction model;

[0021] The risk assessment calculation unit calculates the pest risk value R by using a linear regression model;

[0022] The pest risk value R is obtained by the following formula:

[0023]

[0024] In the formula, β0 represents the intercept term, β1, β2 and β3 represent the processed meteorological data M S , processed biological data B S And the treated pest data PI S The regression coefficient, A t represents the plant growth status at time t, A t-1 Indicates the plant growth status at time t-1.

[0025] By using more flexible expressions, the risk assessment module can more comprehensively capture the multidimensional characteristics of pest and disease risks, improving the accuracy and interpretability of predictions; this provides a more scientific basis for agricultural management decisions and helps farmers and managers take targeted prevention and control measures.

[0026] Preferably, the prediction model module includes a feature extraction and selection unit and a model prediction unit;

[0027] The feature extraction and selection unit includes feature extraction and feature selection, wherein the feature extraction is performed from the processed meteorological data M S and processed biological data B S Extract input features, including air temperature T, air humidity H, rainfall O, plant growth status A and soil moisture F, to form a feature vector X;

[0028] The feature selection collects the pest risk value R and the processed pest data PI from the risk assessment module S , and fit it with the feature vector X, wherein the fitting includes using the information gain measurement method to perform feature selection and construct a feature set X S;

[0029] The feature set X S Fitted by the following formula:

[0030] X S =[R,T,H,O,A,F,PI S ];

[0031] The model prediction unit uses the random forest model to predict the feature set X S The random forest model is trained by constructing multiple decision trees. Each tree is trained by randomly selecting features and samples, and finally the probability P of pest occurrence is calculated and predicted.

[0032] The predicted probability P of pest occurrence is obtained by the following formula:

[0033]

[0034] In the formula, N represents the number of decision trees, h i Represents the i-th tree for the feature set X S The prediction result, mi represents the number of sub-models participating in the voting in the i-th tree, β j represents the weight coefficient, h ij represents the prediction function of the jth node in the i-th tree, e x The random forest model can more comprehensively capture the voting mechanism of different decision trees and their contribution to the final probability prediction, and improve the accuracy and interpretability of the prediction of the probability of pests and diseases. This provides a more scientific basis for agricultural management decisions and helps managers respond to pest and disease risks in a timely manner.

[0035] Preferably, the decision support module includes a risk assessment result processing unit and a comprehensive score calculation and suggestion generation unit;

[0036] The risk assessment result processing unit receives the predicted pest occurrence probability P and related information from the prediction model module, including crop type, pest type m C ={1,2,...,m c} and the cost of pesticide application, and the weight value ω is calculated by the random forest model i ; and provide necessary input for the calculation of comprehensive scores;

[0037] Fit the predicted probability of occurrence of pests and diseases P and related information to obtain the pest and disease type m C The probability of occurrence and constructing the probability feature set P C =[P1,P2,...,P C ];

[0038] Where PC represents the probability of occurrence of the cth pest or disease.

[0039] Preferably, the comprehensive score calculation and suggestion generation unit is based on the probability feature set P X and weight value ω i Calculate the comprehensive score S and generate pesticide application and management recommendations;

[0040] The comprehensive score S is obtained by the following formula:

[0041]

[0042] In the formula, ω i represents the weight value of the i-th feature, P C represents the probability of occurrence of the cth pest;

[0043] The higher the value of the comprehensive score S, the greater the risk of pests and diseases and the higher the management requirements;

[0044] Provide application and management recommendations based on the comprehensive score S, so that the comprehensive score S can be compared with the regulatory threshold ST;

[0045] When the comprehensive score S>regulatory threshold ST, pesticide application is recommended;

[0046] When the comprehensive score S ≤ regulatory threshold ST, monitoring and maintenance are recommended and no pesticide application is required.

[0047] By calculating comprehensive scores and generating pesticide application recommendations, the decision support module can provide a scientific decision-making basis for ecological agricultural management, helping farmers effectively deal with pest and disease risks, optimize resource utilization, and improve crop yield and quality.

[0048] Preferably, the feedback and adjustment module includes an effect data collection unit and a model parameter adjustment unit;

[0049] The effect data collection unit collects effect data ES after the implementation of pesticide application and management measures, and performs model evaluation and parameter adjustment, and the effect data ES includes the proportion of pests and diseases and crop yield;

[0050] Fit the effect data ES into the effect data set E = [e1, e2, ..., e D ];

[0051] In the formula, e D represents the effect data collected after the Dth application;

[0052] The effect data set E is stored to form a database D:

[0053]

[0054] In the formula, D represents the database, which contains all the pesticide application schemes and corresponding effect data, X J E is the feature vector representing the dosing regimen; J Indicates the result data corresponding to the drug administration plan.

[0055] Preferably, the model parameter adjustment unit collects the effect data of the effect data collection unit to analyze the model performance, and adjusts the model parameters, evaluates the deviation ΔPE between the prediction accuracy of the model and the actual effect, adjusts the model parameters, improves the prediction accuracy, and forms a new model;

[0056] The deviation ΔPE is obtained by the following formula:

[0057]

[0058] Where P i represents the predicted probability of occurrence of the i-th pest, E i represents the actual occurrence probability of the i-th pest;

[0059] Adjust the model parameters according to the deviation ΔPE between the prediction accuracy of the evaluation model and the actual effect;

[0060]

[0061] In the formula, represents the updated weight value, represents the old weight value, η represents the learning rate, which controls the step size of parameter update, Represents the gradient of the loss function with respect to the weights.

[0062] Through the implementation of the feedback and adjustment module, the system can continuously collect effect data and optimize model parameters to achieve dynamic adaptation; this dynamic improvement capability enables the agricultural management system to continuously optimize according to actual conditions and improve prediction accuracy, thereby more effectively responding to pest and disease risks and improving crop yield and quality.

[0063] A smart agriculture management method and system applied to ecological agriculture, comprising the following steps:

[0064] Step 1: The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology;

[0065] Step 2: The data preprocessing module performs data cleaning and normalization on the meteorological data M, biological data B and insect pest data PI collected by the data acquisition module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PIS ;

[0066] Step 3: The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R;

[0067] Step 4: The prediction model module uses the random forest machine learning algorithm to predict the risk value R of pests and diseases, and predicts the probability of pests and diseases occurring P;

[0068] Step 5: The decision support module provides pesticide application and management suggestions based on the probability of pests and diseases occurring P, and comprehensively considers ecological agriculture by using the comprehensive score S;

[0069] Step 6: The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

[0070] The present invention provides a smart agricultural management method and system applied to ecological agriculture, which has the following beneficial effects:

[0071] (1) The system can optimize model parameters and feature selection by continuously feeding back predicted results and actual situations. This closed-loop mechanism ensures continuous improvement of the system and further improves the accuracy of predictions and decisions.

[0072] Through refined feature extraction and selection, the model's sensitivity to key influencing factors is improved, and the reliability of prediction results is enhanced. By using the ensemble learning ability of random forests, the model remains efficient in complex environments and reduces dependence on single features or models. The comprehensive scoring system makes pesticide application and management decisions more data-driven, reducing resource waste caused by empirical judgment. The system's dynamic adjustment mechanism reduces unnecessary chemical inputs, helps protect the ecological environment, and promotes the development of ecological agriculture.

[0073] (2) After the model is formed, it can provide a more accurate basis for pesticide application and management decisions. Effective data-driven decision-making can help farmers adjust management measures in a timely manner, reduce resource waste, and increase crop yields. The system continuously updates database D through a feedback mechanism to form a continuously improved data accumulation. This continuous improvement capability helps to refer to historical data in the formulation of future pesticide application plans, optimize pesticide application strategies, and improve overall management efficiency.

[0074] By establishing an effect data set, we can improve our overall understanding of the effects of pesticide application measures, provide data support for decision-making, adjust model parameters in real time, ensure that the model remains efficient in a changing agricultural environment, and enhance the adaptability of the system. The scientific decision-making mechanism based on feedback reduces the waste of resources caused by wrong decisions and improves the economic benefits of crops. By accumulating and analyzing historical data, we can provide more accurate references for future pesticide application plans and promote the intelligence of agricultural management.

[0075] (3) Through the application of a variety of sensors and monitoring equipment, comprehensive collection of meteorological, biological and pest data is achieved, ensuring the integrity and accuracy of the information. The data preprocessing step removes outliers and missing data through cleaning and normalization, improves the reliability of the data, and provides a solid foundation for subsequent analysis.

[0076] The risk assessment module can quickly calculate the risk value R of pests and diseases, identify potential risks in a timely manner, and enable farmers to take early intervention measures to effectively reduce losses. The prediction module using the random forest algorithm can efficiently calculate the probability of pests and diseases P, enhance the ability to predict future risks, and improve the scientific nature of decision-making. The decision support module provides targeted suggestions for pesticide application and management based on the comprehensive score S, helping farmers to allocate resources more reasonably and reduce unnecessary investment. The feedback and adjustment module collects implementation effect data, evaluates and optimizes model parameters in real time, ensures that the system continues to adapt to changes in the agricultural environment, and improves overall management efficiency.

[0077] (4) The feedback and adjustment mechanism enables the model to adjust parameters according to the implementation effect data and achieve dynamic adaptation. Through continuous optimization and adjustment, the system can maintain efficient operation under different climatic and biological conditions and improve the overall agricultural management level. The real-time data collection and processing capabilities enable agricultural management to transform from traditional passive response to active monitoring and early warning, enhancing the ability to predict potential risks.

[0078] Through data cleaning and normalization, the quality of data and the accuracy of analysis are improved, ensuring the scientific nature of risk assessment. The introduction of linear regression model for pest and disease risk assessment makes management decisions more dependent on data support, thereby reducing the impact of human factors. Through dynamic adaptation and continuous improvement, the system can maintain high efficiency in different environments and promote the sustainable development of ecological agriculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 This is a schematic diagram of a flow chart of a smart agricultural management system applied to ecological agriculture in the present invention;

[0080] Figure 2 This is a schematic diagram of the steps of a smart agriculture management method applied to ecological agriculture in the present invention. DETAILED DESCRIPTION

[0081] The technical solutions 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 only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0082] Example 1

[0083] The present invention provides a smart agricultural management method and system applied to ecological agriculture. Figure 1 , including data acquisition module, data preprocessing module, risk assessment module, prediction model module, decision support module and feedback and adjustment module;

[0084] The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology;

[0085] The data preprocessing module performs data cleaning and normalization processing on the meteorological data M, biological data B and insect pest data PI collected by the data collection module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S ;

[0086] The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R;

[0087] The prediction model module uses a random forest machine learning algorithm to perform risk prediction on the pest risk value R and predict the probability of pest occurrence P;

[0088] The decision support module provides pesticide application and management suggestions based on the probability of pest occurrence P, and comprehensively considers ecological agriculture by using the comprehensive score S;

[0089] The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

[0090] In this embodiment, the data acquisition module uses a variety of sensors and monitoring equipment to obtain meteorological data, biological data, and pest data in real time to ensure the timeliness and accuracy of agricultural management decisions. The data preprocessing module cleans and normalizes the collected data, eliminates data noise, improves the accuracy of model input, and provides a solid foundation for subsequent analysis.

[0091] The risk assessment module uses a linear regression model to accurately calculate the risk value R of pests and diseases, allowing farmers to take preventive measures before pests and diseases occur and reduce potential losses. Through the random forest algorithm, the prediction model module can accurately estimate the probability of pests and diseases occurring P, providing a scientific basis for agricultural production and optimizing pesticide application and management strategies.

[0092] The decision support module provides application and management recommendations based on the comprehensive score S, helping farmers make more informed decisions in complex environments and maximize resource utilization efficiency. The feedback and adjustment module collects implementation effect data and adjusts model parameters to ensure that the system can dynamically adapt to changing environments and conditions, improving prediction accuracy and management efficiency.

[0093] Example 2

[0094] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a meteorological data acquisition unit and a biological data and pest data acquisition unit;

[0095] The meteorological data acquisition unit collects meteorological data M including air temperature T, air humidity H and rainfall O through sensors, wherein the sensors include temperature sensors, humidity sensors and rain gauges;

[0096] The biological data and pest data acquisition unit collects biological data B and pest data PI through a plant growth monitor, a high-definition camera, an insect trap and image recognition technology, wherein the biological data B includes plant growth conditions A and soil moisture F, and the plant growth conditions A includes plant height and stem width.

[0097] The data preprocessing module includes a data cleaning unit and a data normalization unit;

[0098] The data cleaning unit cleans the meteorological data M, biological data B and pest data PI by detecting outliers and filling missing values, eliminating outliers and missing values, and obtaining the cleaned meteorological data M. A 、Biological Data B A and Pest Data PI A ;

[0099] The data normalization unit performs the cleaned meteorological data M A 、Biological Data BA and Pest Data PI A Perform normalization processing to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S , making it convenient for subsequent analysis under the same dimension.

[0100] The risk assessment module includes a data input unit and a risk assessment calculation unit;

[0101] The data input unit receives the meteorological data M obtained and processed from the data preprocessing module. S , processed biological data B S And the treated pest data PI S , input into the linear regression prediction model;

[0102] The risk assessment calculation unit calculates the pest risk value R by using a linear regression model;

[0103] The pest risk value R is obtained by the following formula:

[0104]

[0105] In the formula, β0 represents the intercept term, β1, β2 and β3 represent the processed meteorological data M S , processed biological data B S And the treated pest data PI S The regression coefficient, A t represents the plant growth status at time t, A t-1 Indicates the plant growth status at time t-1.

[0106] In this embodiment, through the application of sensors and monitoring equipment, the system can collect meteorological data including air temperature, humidity and rainfall, as well as biological data and pest data in real time. This real-time performance ensures that farmers can respond quickly to environmental changes and take corresponding management measures in a timely manner. Data cleaning and normalization eliminate noise and incomplete data, ensuring the accuracy and consistency of data analysis. This process improves the performance of subsequent risk assessment and prediction models and reduces misjudgments caused by poor data quality.

[0107] The linear regression model is used to calculate the pest risk value R. This method effectively combines meteorological, biological and pest data to provide farmers with scientific risk assessment. This data-driven method can replace traditional empirical judgment and improve the scientificity and accuracy of agricultural management. The decision support module provides farmers with pesticide application and management suggestions based on the pest risk value R, helping them to make comprehensive considerations. This precise management suggestion can effectively reduce the cost of pesticide application and resource waste, and improve the economic benefits of agricultural production.

[0108] Example 3

[0109] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the prediction model module includes a feature extraction and selection unit and a model prediction unit;

[0110] The feature extraction and selection unit includes feature extraction and feature selection, wherein the feature extraction is performed from the processed meteorological data M S and processed biological data B S Extract input features, including air temperature T, air humidity H, rainfall O, plant growth status A and soil moisture F, to form a feature vector X;

[0111] The feature selection collects the pest risk value R and the processed pest data PI from the risk assessment module S , and fit it with the feature vector X, wherein the fitting includes using the information gain measurement method to perform feature selection and construct a feature set X S ;

[0112] The feature set X S Fitted by the following formula:

[0113] X S =[R,T,H,O,A,F,PI S ];

[0114] The model prediction unit uses the random forest model to predict the feature set X S The random forest model is trained by constructing multiple decision trees. Each tree is trained by randomly selecting features and samples, and finally the probability P of pest occurrence is calculated and predicted.

[0115] The predicted probability P of pest occurrence is obtained by the following formula:

[0116]

[0117] In the formula, N represents the number of decision trees, h i Represents the i-th tree for the feature set X S The prediction result, mi represents the number of sub-models participating in the voting in the i-th tree, βj represents the weight coefficient, h ij represents the prediction function of the jth node in the i-th tree, e x Represents the indicator function.

[0118] The decision support module includes a risk assessment result processing unit and a comprehensive score calculation and suggestion generation unit;

[0119] The risk assessment result processing unit receives the predicted pest occurrence probability P and related information from the prediction model module, including crop type, pest type m C ={1,2,...,m c} and the cost of pesticide application, and the weight value ω is calculated by the random forest model i ;

[0120] Fit the predicted probability of occurrence of pests and diseases P and related information to obtain the pest and disease type m C The probability of occurrence and constructing the probability feature set P C =[P1,P2,...,P C ];

[0121] Where P C represents the probability of occurrence of the cth pest or disease.

[0122] The comprehensive score calculation and suggestion generation unit is based on the probability feature set P X and weight value ω i Calculate the comprehensive score S and generate pesticide application and management recommendations;

[0123] The comprehensive score S is obtained by the following formula:

[0124]

[0125] In the formula, ω i represents the weight value of the i-th feature, P C represents the probability of occurrence of the cth pest;

[0126] Provide application and management recommendations based on the comprehensive score S, so that the comprehensive score S can be compared with the regulatory threshold ST;

[0127] When the comprehensive score S>regulatory threshold ST, pesticide application is recommended;

[0128] When the comprehensive score S ≤ regulatory threshold ST, monitoring and maintenance are recommended and no pesticide application is required.

[0129] In this embodiment, key features are extracted from the processed meteorological data and biological data through the feature extraction selection unit to ensure that the input features used in model training are more relevant and effective. This precise extraction helps to improve the accuracy and stability of the prediction model. The random forest model can effectively process high-dimensional data and reduce the risk of overfitting through ensemble learning of multiple decision trees, thereby improving the prediction accuracy of the probability P of pests and diseases. This method makes the system more adaptable and robust in the face of complex agricultural environments.

[0130] The comprehensive score calculation and recommendation generation unit generates scientific pesticide application and management recommendations by combining the predicted probability of pests and diseases with relevant information. This decision support based on data analysis can effectively reduce the cost of pesticide application, improve resource utilization efficiency, and help farmers make better management decisions. By comparing the comprehensive score S with the regulatory threshold ST, the system can dynamically adjust the pesticide application and management recommendations to ensure that the use of pesticides is optimized and the environmental burden is reduced while the risks are controllable. This mechanism promotes the development of sustainable agriculture.

[0131] Example 4

[0132] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the feedback and adjustment module includes an effect data collection unit and a model parameter adjustment unit;

[0133] The effect data collection unit collects effect data ES after the implementation of pesticide application and management measures, and performs model evaluation and parameter adjustment, and the effect data ES includes the proportion of pests and diseases and crop yield;

[0134] Fit the effect data ES into the effect data set E = [e1, e2, ..., e D ];

[0135] In the formula, e D represents the effect data collected after the Dth application;

[0136] The effect data set E is stored to form a database D:

[0137]

[0138] In the formula, D represents the database, which contains all the pesticide application schemes and corresponding effect data, X J E is the feature vector representing the dosing regimen; J Indicates the result data corresponding to the drug administration plan.

[0139] The model parameter adjustment unit collects the effect data of the effect data collection unit to analyze the model performance, and adjusts the model parameters, evaluates the deviation ΔPE between the prediction accuracy of the model and the actual effect, adjusts the model parameters, improves the prediction accuracy, and forms a new model;

[0140] The deviation ΔPE is obtained by the following formula:

[0141]

[0142] Where P i represents the predicted probability of occurrence of the i-th pest, E i represents the actual occurrence probability of the i-th pest;

[0143] Adjust the model parameters according to the deviation ΔPE between the prediction accuracy of the evaluation model and the actual effect;

[0144]

[0145] In the formula, represents the updated weight value, represents the old weight value, η represents the learning rate, which controls the step size of parameter update, Represents the gradient of the loss function with respect to the weights.

[0146] In this embodiment, the effect data collection unit can systematically collect effect data after application and management, including pest and disease ratio and crop yield, and organize them into effect data set E. This systematic monitoring method not only improves the reliability of the data, but also provides a solid foundation for subsequent model evaluation and optimization. The model parameter adjustment unit analyzes the model performance based on the collected effect data and evaluates the deviation between the prediction accuracy and the actual effect. Through this dynamic adjustment mechanism, the model parameters can be continuously optimized to ensure that the model still maintains a high prediction accuracy when facing a complex agricultural environment.

[0147] By calculating the deviation between the predicted results and the actual results, the system can adjust the model parameters in a targeted manner. This process greatly reduces the prediction error caused by model inadaptability or data deviation, and improves the adaptability and practicality of the model.

[0148] Example 5

[0149] A smart agricultural management method applied to ecological agriculture, please refer to Figure 2 , specifically: including the following steps:

[0150] Step 1: The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology;

[0151] Step 2: The data preprocessing module performs data cleaning and normalization on the meteorological data M, biological data B and insect pest data PI collected by the data acquisition module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S ;

[0152] Step 3: The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R;

[0153] Step 4: The prediction model module uses the random forest machine learning algorithm to predict the risk value R of pests and diseases, and predicts the probability of pests and diseases occurring P;

[0154] Step 5: The decision support module provides pesticide application and management suggestions based on the probability of pests and diseases occurring P, and comprehensively considers ecological agriculture by using the comprehensive score S;

[0155] Step 6: The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

[0156] In this embodiment, through steps 1 to 6: through the application of various sensors and monitoring equipment, comprehensive collection of meteorological, biological and pest data is achieved to ensure the integrity and accuracy of the information. The data preprocessing step removes outliers and missing data through cleaning and normalization, improves the reliability of the data, and provides a solid foundation for subsequent analysis.

[0157] The risk assessment module can quickly calculate the risk value R of pests and diseases, identify potential risks in a timely manner, and enable farmers to take early intervention measures to effectively reduce losses. The prediction module using the random forest algorithm can efficiently calculate the probability of pests and diseases P, enhance the ability to predict future risks, and improve the scientific nature of decision-making. The decision support module provides targeted suggestions for pesticide application and management based on the comprehensive score S, helping farmers to allocate resources more reasonably and reduce unnecessary investment. The feedback and adjustment module collects implementation effect data, evaluates and optimizes model parameters in real time, ensures that the system continues to adapt to changes in the agricultural environment, and improves overall management efficiency.

[0158] This method integrates data collection, preprocessing, risk assessment, prediction, decision support and feedback adjustment into one through systematic steps, thus improving the intelligent level of agricultural management. Through scientific data analysis and prediction, it can significantly improve the ability to prevent and control pests and diseases, reduce the use of chemical pesticides, and promote the development of ecological agriculture. Providing scientific decision support based on data avoids the blindness of traditional experience-based decision-making and helps achieve efficient and sustainable agricultural production. The feedback mechanism enables the system to continuously learn and improve, ensuring efficient and accurate management capabilities in a dynamic agricultural environment.

[0159] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart agricultural management system applied to ecological agriculture, characterized by: It includes data collection module, data preprocessing module, risk assessment module, prediction model module, decision support module and feedback and adjustment module; The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology; The data preprocessing module performs data cleaning and normalization processing on the meteorological data M, biological data B and insect pest data PI collected by the data collection module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S ; The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R; The prediction model module uses a random forest machine learning algorithm to perform risk prediction on the pest risk value R and predict the probability of pest occurrence P; The prediction model module includes a feature extraction and selection unit and a model prediction unit; The feature extraction and selection unit includes feature extraction and feature selection, wherein the feature extraction is performed from the processed meteorological data M S and processed biological data B S Extract input features, including air temperature T, air humidity H, rainfall O, plant growth status A and soil moisture F, to form a feature vector X; The feature selection collects the pest risk value R and the processed pest data PI from the risk assessment module S , and fit it with the feature vector X, wherein the fitting includes using the information gain measurement method to perform feature selection and construct a feature set X S ; The feature set X S Fitted by the following formula: X S =[R,T,H,O,A,F,PI S ]; The model prediction unit uses the random forest model to predict the feature set X S The random forest model is trained by constructing multiple decision trees. Each tree is trained by randomly selecting features and samples, and finally the probability P of pest occurrence is calculated and predicted. The predicted probability P of pest occurrence is obtained by the following formula: In the formula, N represents the number of decision trees, h i Represents the i-th tree for the feature set X S The prediction result of , mi represents the number of sub-models participating in the voting in the i-th tree, β j represents the weight coefficient, h ij represents the prediction function of the jth node in the i-th tree, e x represents the indicator function; The decision support module provides pesticide application and management suggestions based on the probability of pest occurrence P, and comprehensively considers ecological agriculture by using the comprehensive score S; The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

2. According to claim 1, a smart agricultural management system applied to ecological agriculture is characterized by: The data collection module includes a meteorological data collection unit and a biological data and pest data collection unit; The meteorological data acquisition unit collects meteorological data M including air temperature T, air humidity H and rainfall O through sensors, wherein the sensors include temperature sensors, humidity sensors and rain gauges; The biological data and pest data acquisition unit collects biological data B and pest data PI through a plant growth monitor, a high-definition camera, an insect trap and image recognition technology, wherein the biological data B includes plant growth conditions A and soil moisture F, and the plant growth conditions A includes plant height and stem width.

3. According to claim 2, a smart agricultural management system applied to ecological agriculture is characterized by: The data preprocessing module includes a data cleaning unit and a data normalization unit; The data cleaning unit cleans the meteorological data M, biological data B and pest data PI by detecting outliers and filling missing values, eliminating outliers and missing values, and obtaining the cleaned meteorological data M. A 、Biological Data B A and Pest Data PI A ; The data normalization unit performs the cleaned meteorological data M A 、Biological Data B A and Pest Data PI A Perform normalization processing to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S , making it convenient for subsequent analysis under the same dimension.

4. The smart agricultural management system applied to ecological agriculture according to claim 3 is characterized by: The risk assessment module includes a data input unit and a risk assessment calculation unit; The data input unit receives the meteorological data M obtained and processed from the data preprocessing module. S , processed biological data B S And the treated pest data PI S , input into the linear regression prediction model; The risk assessment calculation unit calculates the pest risk value R by using a linear regression model; The pest risk value R is obtained by the following formula: In the formula, β0 represents the intercept term, β1, β2 and β3 represent the processed meteorological data M S , processed biological data B S And the treated pest data PI S The regression coefficient, A t represents the plant growth status at time t, A t-1 Indicates the plant growth status at time t-1.

5. The smart agricultural management system applied to ecological agriculture according to claim 1, characterized in that: The decision support module includes a risk assessment result processing unit and a comprehensive score calculation and suggestion generation unit; The risk assessment result processing unit receives the predicted pest occurrence probability P and related information from the prediction model module, including crop type, pest type m C ={1,2,...,m c } and the cost of pesticide application, and the weight value ω is calculated by the random forest model i ; Fit the predicted probability of occurrence of pests and diseases P and related information to obtain the pest and disease type m C The probability of occurrence and constructing the probability feature set P C =[P1,P2,...,P C ]; Where P C represents the probability of occurrence of the cth pest or disease.

6. The smart agricultural management system applied to ecological agriculture according to claim 5, characterized in that: The comprehensive score calculation and suggestion generation unit is based on the probability feature set P X and weight value ω i Calculate the comprehensive score S and generate pesticide application and management recommendations; The comprehensive score S is obtained by the following formula: In the formula, ω i represents the weight value of the i-th feature, P C represents the probability of occurrence of the cth pest; Provide application and management recommendations based on the comprehensive score S, so that the comprehensive score S can be compared with the regulatory threshold ST; When the comprehensive score S>regulatory threshold ST, pesticide application is recommended; When the comprehensive score S ≤ regulatory threshold ST, monitoring and maintenance are recommended and no pesticide application is required.

7. The smart agricultural management system applied to ecological agriculture according to claim 1, characterized in that: The feedback and adjustment module includes an effect data collection unit and a model parameter adjustment unit; The effect data collection unit collects effect data ES after the implementation of pesticide application and management measures, and performs model evaluation and parameter adjustment, and the effect data ES includes the proportion of pests and diseases and crop yield; Fit the effect data ES into the effect data set E = [e1, e2, ..., e D ]; In the formula, e D represents the effect data collected after the Dth application; The effect data set E is stored to form a database D: In the formula, D represents the database, which contains all the application schemes and corresponding effect data, X J E is the feature vector representing the dosing regimen; J Indicates the result data corresponding to the drug administration plan.

8. The smart agricultural management system applied to ecological agriculture according to claim 7, characterized in that: The model parameter adjustment unit collects the effect data of the effect data collection unit to analyze the model performance, and adjusts the model parameters, evaluates the deviation ΔPE between the prediction accuracy of the model and the actual effect, adjusts the model parameters, improves the prediction accuracy, and forms a new model; The deviation ΔPE is obtained by the following formula: Where P i represents the predicted probability of occurrence of the i-th pest, E i represents the actual occurrence probability of the i-th pest; Adjust the model parameters according to the deviation ΔPE between the prediction accuracy of the evaluation model and the actual effect; In the formula, represents the updated weight value, represents the old weight value, η represents the learning rate, which controls the step size of parameter update, Represents the gradient of the loss function with respect to the weights.

9. A smart agricultural management system applied to ecological agriculture, comprising a smart agricultural management method applied to ecological agriculture as described in any one of claims 1 to 8, characterized in that: The following steps are involved: Step 1: The data acquisition module collects meteorological data M through sensors, collects biological data B through plant growth monitors and high-definition cameras, and obtains pest data PI through insect traps and image recognition technology; Step 2: The data preprocessing module performs data cleaning and normalization on the meteorological data M, biological data B and insect pest data PI collected by the data acquisition module to obtain the processed meteorological data M S , processed biological data B S And the treated pest data PI S ; Step 3: The risk assessment module processes the meteorological data M S , processed biological data B S And the treated pest data PI S Input the linear regression prediction model, evaluate the risk of pests and diseases, and calculate the pest and disease risk value R; Step 4: The prediction model module uses the random forest machine learning algorithm to predict the risk value R of pests and diseases, and predicts the probability of pests and diseases occurring P; Step 5: The decision support module provides pesticide application and management suggestions based on the probability of pests and diseases occurring P, and comprehensively considers ecological agriculture by using the comprehensive score S; Step 6: The feedback and adjustment module collects the effect data ES after implementation, adjusts the model parameters, and improves the prediction accuracy.

Images