Crop disease and insect pest intelligent diagnosis system based on One Health and ecological prevention and control method

Abstract

The invention discloses a One Health-based crop disease and insect pest intelligent diagnosis system, which comprises a data acquisition module, a diagnosis prediction module, a prevention and control module, a data storage module and a man-machine interaction module, and comprises the steps of 1, acquiring a crop canopy image and farmland three-dimensional point cloud data, 2, carrying out in-situ monitoring and establishing a soil health index model, step 3, predicting the probability of disease and pest outbreak, step 4, triggering a corresponding prevention and control strategy according to a prediction result, and step 5, recording a prevention and control process and generating a traceability data chain. According to the method, a plant phenotype-microbiome-environmental factor ternary analysis framework is put forward for the first time, multi-source heterogeneous data feature level fusion is performed through a Transform model, cross-domain data deep fusion is realized, the diagnosis accuracy is effectively improved, meanwhile, a deep reinforcement learning adaptive learning mechanism is introduced into the system, a classifier is dynamically updated according to new pest and disease damage data, and the diagnosis accuracy is improved. And the manual annotation demand is effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of crop disease treatment technology, and in particular to a smart diagnostic system and ecological control method for crop diseases and pests based on One Health. Background Technology

[0002] Currently, crop pest and disease control faces three core challenges. First, traditional systems primarily rely on visible light images to identify pests and diseases, lacking comprehensive analysis of soil microbiomes, environmental pollutants, and climatic factors, leading to incomplete diagnostic results. Second, the overuse of chemical pesticides in traditional control methods causes non-point source pollution, damaging soil microbial diversity and indirectly affecting crop disease resistance, creating a vicious cycle. Finally, existing systems often employ independent module designs, failing to form a closed loop between pest and disease early warning, biological control recommendations, and pesticide application optimization, and lacking dynamic learning capabilities. Therefore, this invention proposes a OneHealth-based intelligent diagnosis system and ecological control method for crop pests and diseases to address the problems existing in current technologies. Summary of the Invention

[0003] To address the aforementioned problems, the present invention aims to propose an intelligent diagnostic system and ecological control method for crop diseases and pests based on One Health. This intelligent diagnostic system and ecological control method for crop diseases and pests based on One Health proposes for the first time a ternary analysis framework of plant phenotype-microbiome-environmental factors. Through the Transformer model, feature-level fusion of multi-source heterogeneous data is achieved to realize deep cross-domain data fusion, effectively improving the accuracy of diagnosis. At the same time, the system introduces a deep reinforcement learning adaptive learning mechanism to dynamically update the classifier based on new disease and pest data, effectively reducing the need for manual annotation.

[0004] To achieve the objectives of this invention, the following technical solution is provided: a crop pest and disease intelligent diagnosis system based on One Health, comprising a data acquisition module, a diagnosis and prediction module, a control module, a data storage module, and a human-computer interaction module. The data acquisition module is used to collect remote sensing data on pest and disease migration, plant leaf images, and soil environmental information. The diagnosis and prediction module is used to fuse and analyze the remote sensing data on pest and disease migration, plant leaf images, and soil environmental information to obtain pest diagnosis and prediction results. The control module is used to formulate ecological control plans based on the pest diagnosis and prediction results. The data storage module is used to store historical pest and disease data and a control knowledge base. The human-computer interaction module is used for remote expert diagnosis and intervention and the issuance of control commands.

[0005] Further improvements are made in that: the data acquisition module includes a leaf disease image acquisition unit, a soil environment information acquisition unit, and a remote sensing image acquisition unit. The leaf disease image acquisition unit uses a multispectral camera array to capture crop leaf images on the target plot. The soil environment information acquisition unit collects corresponding soil information of the same target plot based on an integrated sensor array. The remote sensing image acquisition unit is used to collect meteorological satellite data to characterize the migration path of regional diseases and pests.

[0006] Further improvements include: the multispectral camera array includes visible light, near-infrared, and hyperspectral band sensors; the integrated sensor array includes a soil electrochemical sensor array, a soil respiration rate sensor, and a subsurface insect monitoring radar; the soil electrochemical sensor array includes a pH sensor, a conductivity sensor, a temperature and humidity sensor, and a nutrient sensor; and the remote sensing image acquisition unit is equipped with a Fengyun meteorological satellite receiving terminal to analyze MODIS / EVI vegetation index, LST land surface temperature, and NDVI normalized difference vegetation index data.

[0007] Further improvements include: the multispectral camera array is equipped with an automatic exposure compensation algorithm and a cloud and fog removal filter, and adopts a honeycomb deployment structure. The spacing between adjacent nodes is adjusted according to the crop canopy height at a ratio of H / 3, where H is the average crop height; the integrated sensor array forms a self-organizing network through the LoRaWAN protocol; the remote sensing image acquisition unit integrates a Sentinel-2 MSI data source to perform daily updates on 10-meter resolution images.

[0008] Further improvements are made in that the diagnostic prediction module includes a data preprocessing unit, a multimodal feature fusion unit, a regional pest migration path prediction unit, and a dynamic prediction unit. The data preprocessing unit is used to preprocess plant leaf images and soil environmental information to obtain target data. The multimodal feature fusion unit is used to extract corresponding multimodal feature information from the target data and fuse them to obtain fused features. The regional pest migration path prediction unit is based on an improved SEIR infectious disease model and predicts pest migration paths based on pest migration remote sensing data. The dynamic prediction unit is based on an LSTM network and combines fused features and pest migration paths to predict the dynamic risk of pests.

[0009] Further improvements are made in the following aspects: the preprocessing includes image denoising, geometric correction, radiometric calibration, and leaf lesion segmentation; the feature fusion is based on a Transformer encoder with cross-modal attention mechanism to perform three-dimensional feature fusion of spectral features, texture features, and morphological features; and the improved SEIR infectious disease model introduces wind speed, humidity, and temperature meteorological factors as propagation coefficient correction terms when predicting the migration path of pests, where the propagation coefficient β(t) = β0 × exp(α1 × humidity + α2 × temperature + α3 × wind speed), β0 is the basic propagation coefficient, α1 is the influence weight of humidity, α2 is the influence weight of temperature, and α3 is the influence weight of wind speed.

[0010] Further improvements are made in the following aspects: The control module includes an ecological regulation unit, a precision application unit, and an agronomic regulation unit. The ecological regulation unit regulates the crop environment through a configured natural enemy insect release structure, a pheromone dispenser, and a biological pesticide atomization spraying structure. The precision application unit performs grid-based precision application based on a multi-rotor drone swarm equipped with millimeter-wave radar and a variable-speed spray controller. The agronomic regulation unit regulates the balance of crop growth nutrients based on intelligent irrigation pipelines and variable-speed fertilization equipment.

[0011] An ecological control method for crop diseases and pests based on the One Health intelligent diagnostic system includes the following steps:

[0012] Step 1: Acquire crop canopy images using a multispectral camera array and obtain 3D point cloud data of farmland using UAV aerial surveying;

[0013] Step 2: Use an integrated sensor array for in-situ monitoring and combine it with underground insect infestation radar data to establish a soil health index model;

[0014] Step 3: Integrate remote sensing EVI index, soil moisture content and insect population density data, and input them into the improved SEIR model to predict the probability of pest and disease outbreaks;

[0015] Step 4: Trigger corresponding prevention and control strategies based on the prediction results, including activating the emergency prevention and control plan when the risk level is ≥70% and implementing ecological regulation measures when it is ≤30%.

[0016] Step 5: Record the prevention and control process through a blockchain evidence storage system to generate a traceability data chain.

[0017] The further improvement lies in the following: the improved SEIR model in step two includes a susceptible crop area function S(t) affected by the crop growth stage; an exposed crop area function E(t) related to the migration trajectory of vector insects; an infected crop area function I(t) determined based on machine learning prediction results; and a recovery crop area function R(t) considering natural decay and the effects of human intervention.

[0018] The beneficial effects of this invention are as follows: This invention proposes for the first time a ternary analysis framework of plant phenotype-microbiome-environmental factors. Through the Transformer model, feature-level fusion of multi-source heterogeneous data is achieved, realizing deep fusion of cross-domain data and effectively improving the accuracy of diagnosis. At the same time, the system introduces a deep reinforcement learning adaptive learning mechanism to dynamically update the classifier based on new pest and disease data, effectively reducing the need for manual annotation. Furthermore, the system adopts a collaborative control technology of biological control and intelligent pesticide application, which can simultaneously reduce non-point source pollution and pest and disease losses. Attached Figure Description

[0019] Figure 1 This is a diagram of the prevention and control system architecture of the present invention.

[0020] Figure 2 This is a flowchart of the prevention and control method of the present invention. Detailed Implementation

[0021] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0022] Currently, crop disease and pest control faces three core issues:

[0023] Single data dimension: Traditional systems mainly rely on visible light images to identify pests and diseases, lacking comprehensive analysis of soil microbiome, environmental pollutants (such as heavy metals and pesticide residues) and climate factors, resulting in one-sided diagnostic results.37 For example, identifying only leaf lesions may overlook systemic diseases caused by soil microbial imbalance.5

[0024] High ecological risks: Overuse of chemical pesticides leads to non-point source pollution, damages soil microbial diversity, indirectly affects crop disease resistance, and creates a vicious cycle; according to data from the Heilongjiang Plant Protection Station, the utilization rate of pesticides using traditional application methods is less than 40%, and it also leads to a 15% decrease in soil organic matter.

[0025] Poor technical synergy: Existing systems mostly adopt independent module designs, and functions such as pest and disease early warning, biological control recommendation, and pesticide application optimization have not formed a closed loop, and lack dynamic learning capabilities; although the Beijing-Tianjin-Hebei smart plant protection system integrates AI recognition, it does not include microbiome data, and the early warning accuracy rate is only 65%-80%.

[0026] Based on the above issues, according to Figure 1 and Figure 2 As shown, this embodiment provides a crop disease and pest intelligent diagnosis system based on One Health, including a data acquisition module, a diagnosis and prediction module, a prevention and control module, a data storage module, and a human-computer interaction module.

[0027] The data acquisition module is used to collect remote sensing data on the migration of pests and diseases, plant leaf images, and soil environmental information. The data acquisition module includes a leaf disease image acquisition unit, a soil environmental information acquisition unit, and a remote sensing image acquisition unit.

[0028] The leaf disease and pest image acquisition unit uses a multispectral camera array to capture images of crop leaves on the target plot. The multispectral camera array includes sensors for visible light, near-infrared, and hyperspectral bands, supporting centimeter-level spatial resolution and μmol / m 2 • Photosynthetically active radiation accuracy; The multispectral camera array is equipped with an automatic exposure compensation algorithm and a cloud removal filter. The overall deployment structure is honeycomb-shaped, and the spacing between adjacent nodes is adjusted according to the crop canopy height at a ratio of H / 3, where H is the average crop height.

[0029] The soil environmental information acquisition unit collects corresponding soil information of the same target plot based on an integrated sensor array. The integrated sensor array includes a soil electrochemical sensor array, a soil respiration rate sensor, and a ground insect monitoring radar. The integrated sensor array forms a self-organizing network through the LoRaWAN protocol, supports operation in a wide temperature range of -40℃ to 85℃, and has soil water potential measurement function. The soil electrochemical sensor array includes a pH sensor, a conductivity sensor, a temperature and humidity sensor, and a nutrient sensor.

[0030] The remote sensing image acquisition unit is used to collect data from Fengyun meteorological satellites to characterize the migration paths of regional pests and diseases. The remote sensing image acquisition unit is equipped with a Fengyun meteorological satellite receiving terminal to analyze MODIS / EVI vegetation index, LST land surface temperature and NDVI normalized vegetation index data. The remote sensing image acquisition unit integrates Sentinel-2 MSI data source to update 10-meter resolution images daily.

[0031] The diagnosis and prediction module is used to fuse and analyze remote sensing data on the migration of pests and diseases, plant leaf images and soil environmental information to obtain pest diagnosis and prediction results. The diagnosis and prediction module includes a data preprocessing unit, a multimodal feature fusion unit, a regional pest and disease migration path prediction unit and a dynamic prediction unit.

[0032] The data preprocessing unit is used to preprocess plant leaf images and soil environmental information to obtain target data. The preprocessing includes image denoising, geometric correction, radiometric calibration, and leaf lesion segmentation. Specifically, image denoising is achieved through median filtering and BM3D in collaboration, geometric correction is achieved through the RPC model, radiometric calibration is achieved through DN→reflectivity conversion, and leaf lesion segmentation is achieved through the U-Net++ semantic segmentation network.

[0033] The multimodal feature fusion unit is used to extract corresponding multimodal feature information from the target data and fuse them to obtain fused features; feature fusion is a three-dimensional feature fusion of spectral features, texture features and morphological features based on the Transformer encoder with cross-modal attention mechanism;

[0034] The regional pest migration path prediction unit is based on the improved SEIR infectious disease model and predicts pest migration paths based on pest migration remote sensing data. When the improved SEIR infectious disease model predicts pest migration paths, wind speed, humidity and temperature meteorological factors are introduced as correction terms for the propagation coefficient, where the propagation coefficient β(t) = β0 × exp(α1 × humidity + α2 × temperature + α3 × wind speed), β0 is the basic propagation coefficient, α1 is the weight of humidity, α2 is the weight of temperature, and α3 is the weight of wind speed.

[0035] The dynamic prediction unit is based on an LSTM network and combines fused features and pest migration paths to predict the dynamic risk of pests. Specifically, the network input includes multimodal feature vectors, spatial vector data, time series databases, and high-resolution raster data. The output includes dynamic risk heat maps for field visualization and command, risk level zoning for control resource allocation, precision control prescription maps for variable application systems, and decision support reports for agricultural management departments.

[0036] Multimodal feature vectors include spectral features, texture features, and morphological features; spatial vector data includes pest migration direction, speed, and carriers; time-series data includes historical disease coordinates and affected area; and high-resolution raster data includes current crop growth and soil moisture.

[0037] The prevention and control module is used to formulate ecological prevention and control plans based on pest diagnosis and prediction results. The prevention and control module includes an ecological regulation unit, a precision application unit, and an agronomic regulation unit.

[0038] The ecological regulation unit regulates the crop environment through the configuration of natural enemy insect release structures (such as Trichogramma wasps and ladybugs), sex pheromone dispensers, and biological pesticide atomization spraying structures. The precision application unit performs grid-based precision application based on a swarm of multi-rotor drones equipped with millimeter-wave radar and variable spray controllers. The agronomic regulation unit regulates the balance of crop growth nutrients based on intelligent irrigation pipelines and variable-speed fertilization equipment.

[0039] The data storage module is used to store historical pest and disease data and a knowledge base for prevention and control.

[0040] The human-computer interaction module is used for remote diagnosis, intervention, and control command issuance by experts.

[0041] The data transmission network between modules includes: an edge computing gateway, which uses the NVIDIA Jetson Xavier NX platform and runs the TensorRT acceleration engine to achieve localized processing of image feature extraction; a hybrid networking architecture, in which the core backbone network uses a 5G slicing network, eMBB slicing ensures video stream transmission, uRLLC slicing ensures control command transmission, and the edge side uses LoRaWAN networking to cover blind spots in farmland; and a network security mechanism, which deploys a blockchain evidence storage system to hash and encrypt key control commands to prevent data tampering.

[0042] An ecological control method for crop diseases and pests based on the One Health intelligent diagnostic system includes the following steps:

[0043] Step 1: Acquire crop canopy images using a multispectral camera array and obtain 3D point cloud data of farmland using UAV aerial surveying;

[0044] Step 2: Use an integrated sensor array for in-situ monitoring and combine it with underground insect infestation radar data to establish a soil health index model;

[0045] Step 3: Integrate remote sensing EVI index, soil moisture content and insect population density data, and input them into the improved SEIR model to predict the probability of pest and disease outbreaks;

[0046] The improved SEIR model includes a susceptible crop area function S(t) affected by the crop growth stage; an exposed crop area function E(t) related to the migration trajectory of vector insects; an infected crop area function I(t) determined based on machine learning prediction results; and a recovery crop area function R(t) considering natural decay and the effects of human intervention.

[0047] Step 4: Trigger corresponding prevention and control strategies based on the prediction results, including activating the emergency prevention and control plan when the risk level is ≥70% and implementing ecological regulation measures when it is ≤30%.

[0048] Step 5: Record the prevention and control process through a blockchain evidence storage system to generate a traceability data chain.

[0049] When carrying out ecological control, the following work should be done:

[0050] Diagnostic phase: Identify rhizosphere microbial imbalances (such as decreased actinomycete abundance) through microbiome sequencing, and associate them with the suppression of crop disease resistance genes caused by heavy metal pollution.

[0051] Recommended biocontrol method: Match antagonistic bacteria (such as Bacillus subtilis inhibiting rice blast) according to the type of pathogen, and calculate the optimal inoculation concentration (based on a microbial interaction database).

[0052] Precision application optimization: Combining pest and disease severity levels with soil pollution data, generate pesticide reduction plans (such as targeted release via nanocarriers, reducing pesticide dosage by 30%-50%).

[0053] Intervention phase:

[0054] Microbial regulation: Inoculate with PGPR (plant rhizosphere growth promoters) and pathogen antagonists to restore soil microbial balance.

[0055] Ecological barrier construction: Plant vetiver (to absorb heavy metals) and marigold (to attract and kill pests) at the boundaries of farmland to form a biological isolation zone.

[0056] Effect evaluation: Data such as pesticide application rate and changes in microbial diversity are recorded through blockchain to quantify ecological benefits (e.g., 40% reduction in pesticide use and a 3-fold increase in the number of natural enemy insects).

[0057] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart diagnostic system for crop diseases and pests based on One Health, characterized in that: It includes a data acquisition module, a diagnosis and prediction module, a prevention and control module, a data storage module, and a human-computer interaction module. The data acquisition module is used to collect remote sensing data on pest migration, plant leaf images, and soil environmental information. The diagnosis and prediction module is used to fuse and analyze the remote sensing data on pest migration, plant leaf images, and soil environmental information to obtain pest diagnosis and prediction results. The prevention and control module is used to formulate ecological prevention and control plans based on the pest diagnosis and prediction results. The data storage module is used to store historical pest data and a prevention and control knowledge base. The human-computer interaction module is used for remote expert diagnosis and intervention and the issuance of prevention and control instructions.

2. The intelligent crop disease and pest diagnosis system based on One Health according to claim 1, characterized in that: The data acquisition module includes a leaf disease image acquisition unit, a soil environment information acquisition unit, and a remote sensing image acquisition unit. The leaf disease image acquisition unit uses a multispectral camera array to capture crop leaf images on the target plot. The soil environment information acquisition unit collects corresponding soil information of the same target plot based on an integrated sensor array. The remote sensing image acquisition unit is used to collect meteorological satellite data to characterize the migration path of regional diseases and pests.

3. The intelligent crop disease and pest diagnosis system based on One Health according to claim 2, characterized in that: The multispectral camera array includes visible light, near-infrared, and hyperspectral band sensors. The integrated sensor array includes a soil electrochemical sensor array, a soil respiration rate sensor, and a subsurface insect monitoring radar. The soil electrochemical sensor array includes a pH sensor, a conductivity sensor, a temperature and humidity sensor, and a nutrient sensor. The remote sensing image acquisition unit is equipped with a Fengyun meteorological satellite receiving terminal to analyze MODIS / EVI vegetation index, LST land surface temperature, and NDVI normalized difference vegetation index data.

4. The intelligent crop disease and pest diagnosis system based on One Health according to claim 2, characterized in that: The multispectral camera array is equipped with an automatic exposure compensation algorithm and a cloud removal filter. It adopts a honeycomb deployment structure, and the spacing between adjacent nodes is adjusted according to the crop canopy height at a ratio of H / 3, where H is the average crop height. The integrated sensor array forms a self-organizing network through the LoRaWAN protocol. The remote sensing image acquisition unit integrates a Sentinel-2 MSI data source and performs daily updates on 10-meter resolution images.

5. The intelligent crop disease and pest diagnosis system based on One Health according to claim 1, characterized in that: The diagnostic prediction module includes a data preprocessing unit, a multimodal feature fusion unit, a regional pest migration path prediction unit, and a dynamic prediction unit. The data preprocessing unit is used to preprocess plant leaf images and soil environmental information to obtain target data. The multimodal feature fusion unit is used to extract corresponding multimodal feature information from the target data and fuse them to obtain fused features. The regional pest migration path prediction unit predicts pest migration paths based on an improved SEIR infectious disease model and pest migration remote sensing data. The dynamic prediction unit predicts pest dynamic risk based on an LSTM network and combines fused features and pest migration paths.

6. The intelligent crop disease and pest diagnosis system based on One Health according to claim 5, characterized in that: The preprocessing includes image denoising, geometric correction, radiometric calibration, and leaf lesion segmentation. The feature fusion is based on a Transformer encoder with a cross-modal attention mechanism to perform three-dimensional feature fusion of spectral features, texture features, and morphological features. When the improved SEIR infectious disease model predicts the migration path of pests, it introduces wind speed, humidity, and temperature meteorological factors as correction terms for the propagation coefficient, where the propagation coefficient β(t) = β0 × exp(α1 × humidity + α2 × temperature + α3 × wind speed), β0 is the basic propagation coefficient, α1 is the weight of humidity, α2 is the weight of temperature, and α3 is the weight of wind speed.

7. The intelligent crop disease and pest diagnosis system based on One Health according to claim 1, characterized in that: The control module includes an ecological regulation unit, a precision pesticide application unit, and an agronomic regulation unit. The ecological regulation unit regulates the crop environment through a configured natural enemy insect release structure, a pheromone dispenser, and a biological pesticide atomization spraying structure. The precision pesticide application unit performs grid-based precision pesticide application based on a swarm of multi-rotor drones equipped with millimeter-wave radar and a variable-speed spray controller. The agronomic regulation unit regulates the balance of crop growth nutrients based on intelligent irrigation pipelines and variable-speed fertilization equipment.

8. An ecological control method for crop pests and diseases based on a One Health-based intelligent diagnostic system, as described in any one of claims 1-7, characterized in that, Includes the following steps: Step 1: Acquire crop canopy images using a multispectral camera array and obtain 3D point cloud data of farmland using UAV aerial surveying; Step 2: Use an integrated sensor array for in-situ monitoring and combine it with underground insect infestation radar data to establish a soil health index model; Step 3: Integrate remote sensing EVI index, soil moisture content and insect population density data, and input them into the improved SEIR model to predict the probability of pest and disease outbreaks; Step 4: Trigger corresponding prevention and control strategies based on the prediction results, including activating the emergency prevention and control plan when the risk level is ≥70% and implementing ecological regulation measures when it is ≤30%. Step 5: Record the prevention and control process through a blockchain evidence storage system to generate a traceability data chain.

9. The ecological control method for crop pests and diseases based on a One Health-based intelligent diagnostic system, as described in claim 8, is characterized in that: The improved SEIR model in step two includes a susceptible crop area function S(t) affected by the crop growth stage; an exposed crop area function E(t) related to the migration trajectory of vector insects; an infected crop area function I(t) determined based on machine learning prediction results; and a recovery crop area function R(t) considering natural decay and human intervention effects.

Images