Method and system for detecting, diagnosing and regulating plant diseases and insect pests

Abstract

The invention provides a plant disease and insect pest detection-diagnosis-regulation method and system, and relates to the technical field of intelligent agriculture and forestry management. According to the method, multi-source monitoring data such as plant images, soil physicochemistry, plant physiology and environmental meteorology are integrated, comprehensive diagnosis and prediction are carried out based on the AI analysis model, the limitation of single data source analysis is overcome, and the early recognition accuracy and early warning timeliness of plant health conditions and plant disease and insect pest risks in a complex environment are remarkably improved. Through cross validation and complementation of multi-dimensional data, misjudgment and missed judgment are reduced, the diagnosis result is more comprehensive and reliable, and a solid data foundation is laid for precise regulation and control.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture and forestry management technology, and in particular to a plant disease and pest detection-diagnosis-control system and method, as well as a data acquisition device for monitoring plant diseases and pests. Background Technology

[0002] Plant diseases and pests are significant factors affecting agricultural and forestry production. Traditional control methods mainly rely on manual inspections and experience-based judgment, which suffer from low efficiency, poor accuracy, and delayed response. With technological advancements, pest and disease detection systems based on single data sources (such as image recognition) have emerged. However, these systems often suffer from limitations such as limited monitoring dimensions, poor environmental adaptability, and limited prediction accuracy.

[0003] In summary, existing plant disease and pest management systems generally suffer from the following problems: insufficient multi-source data fusion capabilities, making it difficult to comprehensively reflect the plant health status; limited accuracy of disease and pest prediction models, resulting in weak early warning capabilities; disconnect between monitoring and control, failing to form an intelligent closed-loop management system; and insufficient system adaptability and learning capabilities, making it difficult to adapt to complex and ever-changing application environments. Summary of the Invention

[0004] In view of this, the present invention provides a method and system for detecting, diagnosing and regulating plant diseases and pests. The technical problem to be solved is: how to achieve intelligent, precise and closed-loop early monitoring, diagnosis and regulation integrated management of plant diseases and pests; thereby overcoming the problems of single monitoring dimensions, limited prediction accuracy, delayed regulation response and lack of closed-loop management in the existing plant disease and pest management system.

[0005] This invention provides a method for detecting, diagnosing, and regulating plant diseases and pests, comprising the following steps: S1, acquiring multi-source monitoring data on plant growth and diseases and pests, including plant image data, soil physicochemical index data, plant physiological index data, and environmental meteorological data; S2, based on the multi-source monitoring data, diagnosing plant health status and predicting disease and pest risks using an AI analysis model, and generating diagnostic results and early warning information; S3, based on the diagnostic results and early warning information, generating and recommending at least one prevention and control strategy; S4, through an intelligent decision-making and control module, optimizing the recommended prevention and control strategy and generating executable control instructions.

[0006] Optionally, the steps for acquiring multi-source monitoring data include: plant image data and temperature data acquired through the plant image monitoring unit; soil physicochemical index data acquired through the soil and root monitoring unit, including one or more of pH value, water content, electrical conductivity, heavy metal content, root growth status, root oxygen density, and total carbon content around the roots; plant physiological data acquired through the plant index measurement module, including one or more of chlorophyll content and plant growth rate; and meteorological data acquired or retrieved through the meteorological data monitoring unit.

[0007] Optionally, the soil and root monitoring unit includes a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe; the soil volumetric water content data collected by the dielectric constant sensor is calibrated in real time using a temperature compensation algorithm.

[0008] Optionally, the prevention and control strategies generated by the intelligent decision-making and control module include at least one of the following: physical control strategies, biological pesticide application strategies, intelligent chemical pesticide application strategies, ecological agronomic prevention and control strategies, or strategies requesting human intervention.

[0009] Optionally, the AI ​​analysis model includes: an image recognition sub-model for classifying and locating plant images to identify pest and disease types; a disease prediction sub-model for predicting disease epidemic risk based on environmental meteorological data and historical data using long short-term memory networks or extreme gradient boosting algorithms; and / or an insect infestation prediction sub-model for predicting insect population density trends based on insect images and trajectory data.

[0010] This invention also provides a plant disease and pest detection-diagnosis-control system for implementing the above-mentioned method, comprising: an intelligent monitoring and data acquisition module for acquiring multi-source monitoring data of plants through multiple sensors and image acquisition devices deployed in the monitoring area; a data processing and communication module, communicatively connected to the intelligent monitoring and data acquisition module, for receiving and preprocessing the multi-source monitoring data and transmitting the processed data through a communication network; a disease and pest diagnosis and prediction module, including at least one AI analysis model, communicatively connected to the data processing and communication module, which diagnoses the plant health status based on the data transmitted by the data processing and communication module using the AI ​​analysis model, and outputs diagnostic results and early warning information; an intelligent decision-making and control module, communicatively connected to the disease and pest diagnosis and prediction module, which generates at least one targeted prevention and control strategy based on the received diagnostic results and early warning information; and an automated execution interface module, communicatively connected to the intelligent decision-making and control module, which drives one or more external execution devices to perform corresponding physical control operations according to the prevention and control strategy.

[0011] Optionally, the intelligent monitoring and data acquisition module includes: a plant image monitoring unit for acquiring visible light and infrared thermal images of plants via drones or fixed cameras; a soil and root monitoring unit for monitoring soil pH, moisture content, electrical conductivity, heavy metal content, and root growth density and oxygen density via in-situ probes; a plant physiological indicator monitoring unit for monitoring plant chlorophyll content and growth rate; and a meteorological data monitoring unit for acquiring temperature, humidity, light, and precipitation data for the monitored area.

[0012] Optionally, the external execution devices driven by the automated execution interface module include intelligent spraying systems, drones, integrated water and fertilizer equipment, predator release devices, or warning sign triggering devices.

[0013] This invention also provides a data acquisition device for monitoring plant diseases and pests, comprising: a housing; a multi-channel sensor array integrated into the housing or connected to the housing via a cable, the multi-channel sensor array including a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe, for in-situ acquisition of soil volumetric water content, dielectric constant, pH value, conductivity, and heavy metal content data; a data processor built into the housing and communicatively connected to the multi-channel sensor array, the data processor being configured to perform real-time calibration of the data acquired by the dielectric constant sensor using a preset temperature compensation algorithm, and to perform pollutant concentration inversion on the data acquired by the miniature XRF probe using a preset random forest regression model; a memory communicatively connected to the data processor, the memory being used to store the calibrated and inverted data; and a communication interface for transmitting the stored data to an external system.

[0014] Implementing this invention offers the following beneficial effects: By integrating multi-source monitoring data, including plant images, soil physiology and chemistry, plant physiology, and environmental meteorology, and using an AI analysis model for comprehensive diagnosis and prediction, this invention overcomes the limitations of single-data source analysis. It significantly improves the accuracy and timeliness of early identification of plant health status and pest and disease risks in complex environments. Cross-validation and complementarity of multi-dimensional data reduce misjudgments and omissions, resulting in more comprehensive and reliable diagnostic results, laying a solid data foundation for precise regulation. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of the plant disease and pest detection-diagnosis-control method of the present invention in one embodiment.

[0016] Figure 2 This is a schematic diagram of the module connection of the plant disease and pest detection-diagnosis-control system of the present invention in one embodiment. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only for ease of explanation and do not limit the order of the steps. The execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.

[0018] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0019] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0020] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0021] Example 1 In this embodiment, as Figure 1 The method for detecting, diagnosing, and controlling plant diseases and pests, as shown, includes the following steps: S1. Obtain multi-source monitoring data on plant growth and pests and diseases; S2. Based on multi-source monitoring data, AI analysis models are used to diagnose plant health status and predict pest and disease risks, and generate diagnostic results and early warning information. S3. Based on the diagnostic results and early warning information, generate and recommend at least one prevention and control strategy; S4. Through the intelligent decision-making and control module, the recommended prevention and control strategies are optimized, and executable control instructions are generated.

[0022] Specifically, this invention overcomes the limitations of single-source data analysis by integrating multi-source monitoring data such as plant images, soil physiology and chemistry, plant physiology, and environmental meteorology, and using AI analysis models for comprehensive diagnosis and prediction. This significantly improves the accuracy and timeliness of early identification of plant health status and pest and disease risks in complex environments. Cross-validation and complementarity of multi-dimensional data reduce misjudgments and omissions, making the diagnostic results more comprehensive and reliable, thus laying a solid data foundation for precise regulation.

[0023] In this embodiment, the multi-source monitoring data includes plant image data, soil physicochemical index data, plant physiological index data, and environmental meteorological data.

[0024] Specifically, the steps for acquiring multi-source monitoring data include: Plant image data and temperature data acquired through the plant image monitoring unit; Soil physicochemical index data obtained through the soil and root monitoring unit include one or more of the following: pH value, water content, electrical conductivity, heavy metal content, root growth status, root oxygen density, and total carbon content around the roots. Plant physiological data obtained through the plant index measurement module includes one or more of the following: chlorophyll content and plant growth rate. Meteorological data obtained or retrieved through meteorological data monitoring units.

[0025] In this embodiment, the soil and root monitoring unit includes a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe; The soil volumetric water content data collected by the dielectric constant sensor is calibrated in real time using a temperature compensation algorithm.

[0026] In this embodiment, the prevention and control strategies generated by the intelligent decision-making and control module include at least one of the following: physical control strategy, biological pesticide application strategy, intelligent chemical pesticide application strategy, ecological agronomic prevention and control strategy, or strategy requesting human intervention.

[0027] In this embodiment, the AI ​​analysis model includes: An image recognition sub-model is used to classify and locate plant images in order to identify the types of pests and diseases; The disease prediction sub-model is used to predict the risk of disease outbreaks based on environmental meteorological data and historical data, using long short-term memory networks or extreme gradient boosting algorithms. And / or an insect infestation prediction sub-model, used to predict insect population density trends based on insect images and trajectory data.

[0028] Example 2 Based on Embodiment 1, this embodiment provides a plant disease and pest detection-diagnosis-control system for implementing the above method.

[0029] In this embodiment, as Figure 2 The plant disease and pest detection-diagnosis-control system shown includes: The system comprises the following modules: an intelligent monitoring and data acquisition module, used to acquire multi-source monitoring data of plants through multiple sensors and image acquisition devices deployed in the monitoring area; a data processing and communication module, communicatively connected to the intelligent monitoring and data acquisition module, used to receive and preprocess the multi-source monitoring data and transmit the processed data through a communication network; a pest and disease diagnosis and prediction module, including at least one AI analysis model, communicatively connected to the data processing and communication module, which diagnoses plant health status based on the data transmitted by the data processing and communication module using the AI ​​analysis model, and outputs diagnostic results and early warning information; an intelligent decision-making and control module, communicatively connected to the pest and disease diagnosis and prediction module, which generates at least one targeted prevention and control strategy based on the received diagnostic results and early warning information; and an automated execution interface module, communicatively connected to the intelligent decision-making and control module, which drives one or more external execution devices to perform corresponding physical control operations according to the prevention and control strategy.

[0030] In this embodiment, the plant disease and pest detection-diagnosis-control system adopts a layered hybrid communication architecture that is collaborative between the edge and the cloud.

[0031] Specifically, the intelligent monitoring and data acquisition module comprises multiple sensors and image acquisition devices serving as the field perception layer, while the data processing and communication module is deployed within the facility greenhouse and fixed monitoring stations as the edge aggregation layer. The edge aggregation layer provides RS-485 and RJ45 interfaces, as well as a wireless communication module. The edge aggregation layer connects to the field perception layer via wired / short-range high-speed connections, specifically using an RS-485 bus or Ethernet. The automation execution interface module, the field perception layer, and the edge aggregation layer together constitute the edge layer in the edge-cloud layered hybrid communication architecture. The automation execution interface module and the data processing and communication module are connected via NB-IoT or 4G / 5G cellular networks.

[0032] Specifically, the pest and disease diagnosis and prediction module and the intelligent decision-making and control module are both deployed in the cloud as a remote transmission layer, with NB-IoT / 4G communication modules and SIM cards built into the edge gateway. The edge aggregation layer and the remote transmission layer are connected via NB-IoT or 4G / 5G cellular networks.

[0033] Specifically, the multi-source monitoring data includes plant image data, soil physicochemical index data, plant physiological index data, and environmental meteorological data.

[0034] In this embodiment, multi-source monitoring data is mainly collected through a plant image monitoring unit, a soil and root monitoring unit, a plant physiological index measurement unit, and a meteorological data monitoring unit.

[0035] Specifically, the plant image monitoring unit acquires plant image data via drones or fixed cameras, including visible light images and infrared thermal images. Subsequently, an AI analysis module automatically compares the obtained plant image data with ideal growth images of the target plant to determine the plant's growth suitability. Simultaneously, the infrared thermal images can assist in monitoring abnormal plant canopy temperatures, thereby determining whether there is pathogen infection or insect feeding activity.

[0036] In this embodiment, plant roots are crucial for plant growth, and in southern regions, many plants exhibit normal external stem and leaf growth, but their roots are rotting and pose a risk of collapse. Therefore, this embodiment also monitors soil physicochemical indicators using a soil and root monitoring unit.

[0037] Specifically, the soil and root monitoring unit mainly uses in-situ probes to monitor soil pH, moisture content, porosity, electrical conductivity, redox potential, total carbon, heavy metals and other indicators in real time. At the same time, the soil and root monitoring unit mainly measures root growth, root oxygen density and total carbon content around the roots and other indicators. The data collected from both are combined and transmitted to the pest and disease diagnosis and prediction module.

[0038] More specifically, the soil and root monitoring unit integrates multiple in-situ sensor probes to monitor different parameters, including: Dielectric constant sensors (such as TEROS 12) based on the frequency domain reflection (FDR) principle can monitor soil volumetric water content and electrical conductivity in real time. Soil pH was monitored using an ion-selective field-effect transistor (ISFET) pH sensor. A platinum electrode redox potential sensor is used to monitor the soil redox potential; A miniature X-ray fluorescence (XRF) probe is used for in-situ semi-quantitative analysis of heavy metal elements (such as lead, cadmium, and arsenic) in soil. A microroot window (Minirhizotron) imaging system was used to periodically acquire root images, and root growth density and biomass were quantified using image analysis software. A miniature fiber optic oxygen sensor is inserted into the rhizosphere soil to monitor root oxygen density.

[0039] Soil porosity is calculated by the data processor based on moisture sensors and a soil density model. Total soil carbon and total carbon content around the roots are analyzed by periodically collecting rhizosphere soil samples and sending them to the laboratory. The analysis results are then input into the pest and disease diagnosis and prediction module as offline calibration data.

[0040] In this embodiment, plant physiological index data are mainly obtained through the plant physiological index measurement unit. The plant physiological index measurement unit is used to measure indicators such as plant chlorophyll content and plant growth rate to determine the plant growth trend. The plant chlorophyll content, plant growth rate and other indicators are input into the disease and pest diagnosis and prediction module. The AI ​​analysis module determines the plant growth cycle by comparing it with the pre-stored standard data of plant physiological indicators.

[0041] In this embodiment, the meteorological data monitoring unit is mainly used to retrieve meteorological station data to collect data, thereby obtaining environmental meteorological data, and then determining soil moisture, plant growth health parameters, etc., to further screen the optimal growth factors and provide a basis for screening the characteristics of pests and diseases in plants.

[0042] In this embodiment, the AI ​​analysis model includes: An image recognition sub-model is used to classify and locate plant images in order to identify the types of pests and diseases; The disease prediction sub-model is used to predict the risk of disease outbreaks based on environmental meteorological data and historical data, using Long Short-Term Memory (LSTM) or eXtreme Gradient Boosting (XGBoost) algorithms. And / or an insect infestation prediction sub-model, used to predict insect population density trends based on insect images and trajectory data.

[0043] Specifically, the image recognition sub-model mainly compares the data and images collected by the intelligent monitoring and data acquisition module with a pre-stored database containing plant diseases and pests and plant growth status through an AI analysis model. Based on the optimal growth status, the plant growth is divided into four states: healthy, sub-healthy, early stage of disease and pest, and late stage of disease and pest.

[0044] In this embodiment, the disease prediction sub-model uses LSTM and XGBoost algorithms to perform deep learning image recognition, disease occurrence prediction, and insect infestation dynamic prediction.

[0045] Specifically, deep learning image recognition includes automatic classification and localization of acquired images, achieving high-precision identification of common diseases (such as downy mildew, anthracnose, and rust) and pests (such as aphids, spider mites, and whiteflies). Disease occurrence prediction involves combining meteorological data, crop growth stages, and historical disease records, using algorithms such as LSTM or XGBoost to establish a disease epidemic risk prediction model, issuing early warnings 3–7 days in advance. The pest prediction sub-model predicts insect population density trends based on insect-attracting lamp image recognition and flight trajectory analysis, guiding the optimal application window for pesticides. Finally, the disease and pest diagnosis and prediction module assesses the severity of plant diseases and pests based on dynamic monitoring indicators and model predictions, generating diagnostic results and early warning information for potential sudden situations.

[0046] In this embodiment, the intelligent decision-making and control module generates prevention and control strategies based on the diagnostic results and early warning information output by the pest and disease diagnosis and prediction module. The prevention and control strategies include physical control, biological pesticide application strategies, intelligent application of chemical pesticides, ecological agronomic prevention and control strategies, or strategies requesting human intervention.

[0047] Specifically, physical control methods include recommending natural enemy insects (ladybugs, predatory mites), setting up yellow and blue sticky traps, and using insect nets to prevent and remove plant diseases and pests. Intelligent application of chemical pesticides involves matching the types of diseases and pests and recommending suitable microbial agents (such as Bacillus thuringiensis, Bacillus subtilis) or plant-derived pesticides. Intelligent application of chemical pesticides, when necessary, recommends low-toxicity, low-residue pesticides and accurately calculates the application rate and spraying area to avoid overuse. Ecological agronomic control strategies suggest agronomic management methods such as crop rotation, intercropping, adjusting planting density, and improving ventilation and light penetration to enhance crop resistance. The strategy of requesting human intervention refers to the intelligent decision-making and control module indicating the need for human intervention based on root data monitoring obtained from soil and root monitoring units and the severity of plant diseases and pests, indicating a risk of plant collapse. After human intervention, soil and plant indicators are checked one by one, and data is updated and matched in a timely manner, ultimately determining whether to maintain the plant or cut it down. It should be noted that the plant disease and pest detection-diagnosis-control system only provides early warning and data analysis support. All major decisions must be made after on-site confirmation by professionals.

[0048] In this embodiment, the intelligent decision-making and control module includes a multi-source data fusion engine, an intelligent decision-making algorithm, a remote management platform, a self-learning and model retraining pipeline, etc.

[0049] Specifically, the multi-source data fusion engine integrates multi-dimensional heterogeneous data from intelligent monitoring and data acquisition modules, including hyperspectral images, visible light images, infrared thermal images, environmental temperature and humidity, light intensity, soil moisture and weather forecasts, etc., and constructs a digital representation model of crop health status through spatial-temporal alignment and feature extraction algorithms.

[0050] Specifically, the multi-source data fusion engine employs data fusion algorithms, such as Kalman Filter, Dempster-Shafer Theory, or Deep Feature Fusion Network, to improve information utilization and state recognition accuracy, providing a reliable data foundation for subsequent decision-making. In this embodiment, the intelligent decision-making algorithm combines the data fused by the multi-source data fusion engine with diagnostic results and early warning information, and generates prevention and control strategies by combining knowledge graph, rule-based reasoning and machine learning (ML).

[0051] In this embodiment, the plant disease and pest detection-diagnosis-control system incorporates a crop growth model and a pest and disease occurrence model, supporting multi-objective optimization (MOO). MOO objectives include maximizing control effectiveness, minimizing pesticide use, and minimizing ecological impact. Through the synergistic effect of the crop growth model, pest and disease occurrence model, and multi-objective optimization algorithm, personalized recommendations are achieved for each location and each disease. Furthermore, the multi-objective optimization algorithm supports dynamic adjustment, allowing for real-time adjustments to optimization objectives and parameters based on user feedback and execution results, thereby correcting the decision-making path.

[0052] In this embodiment, the remote management platform is a cloud-based visual management platform that supports access from both web and mobile devices and provides multi-role access control (farmers, agricultural technicians, and managers). The visual management platform has functions such as task scheduling, progress monitoring, prevention and control record archiving, effect evaluation, and report generation, enabling full-process traceability management.

[0053] In this embodiment, the self-learning and model iteration mechanism enables the plant disease and pest detection-diagnosis-control system to have continuous learning capabilities, automatically collect newly acquired image data, prevention and control records and field feedback, and regularly retrain and optimize the image recognition model, prediction algorithm and decision rules.

[0054] Specifically, the self-learning and model iteration mechanism introduces incremental learning and federated learning (FL) mechanisms to achieve cross-regional knowledge sharing while protecting data privacy, regularly update the identification model and prediction algorithm, improve long-term applicability, and thus continuously improve the adaptability and intelligence level of the plant disease and pest detection-diagnosis-control system.

[0055] In this embodiment, the automated execution interface module interfaces with the intelligent decision-making and control module and external execution devices according to a standardized communication protocol, enabling the automatic issuance and execution of prevention and control commands. External execution devices include intelligent spraying systems, drones, integrated water and fertilizer systems, natural enemy release devices, or warning sign triggering devices. This allows for variable-rate application and precise application, significantly improving operational efficiency and response speed while reducing the cost of manual intervention.

[0056] In this embodiment, the multi-source monitoring data includes plant image data, soil physicochemical index data, plant physiological index data, and environmental meteorological data.

[0057] In this embodiment, the intelligent monitoring and data acquisition module includes: a plant image monitoring unit, used to acquire visible light images and infrared thermal images of plants via a drone or a fixed camera; a soil and root monitoring unit, used to monitor soil pH, moisture content, electrical conductivity, heavy metal content, and root growth density and oxygen density via an in-situ probe; a plant physiological index monitoring unit, used to monitor plant chlorophyll content and growth rate; and a meteorological data monitoring unit, used to acquire temperature, humidity, light, and precipitation data of the monitored area.

[0058] In this embodiment, the soil and root monitoring unit includes a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe; the data processing and communication module is configured to use a temperature compensation algorithm to calibrate the soil volumetric water content data collected by the dielectric constant sensor in real time.

[0059] In this embodiment, the prevention and control strategies generated by the intelligent decision-making and control module include at least one of the following: physical control strategy, biological pesticide application strategy, intelligent chemical pesticide application strategy, ecological agronomic prevention and control strategy, or strategy requesting human intervention.

[0060] In this embodiment, the external execution devices driven by the automated execution interface module include intelligent spraying systems, drones, integrated water and fertilizer equipment, predator release devices, or warning sign triggering devices.

[0061] Example 3 This embodiment provides a data acquisition device for monitoring plant diseases and pests, based on any of the above embodiments.

[0062] In this embodiment, the data acquisition device for monitoring plant diseases and pests includes: a housing; a multi-channel sensor array integrated into the housing or connected to the housing via a cable, the multi-channel sensor array including a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe, used for in-situ acquisition of soil volumetric water content, dielectric constant, pH value, conductivity, and heavy metal content data; a data processor built into the housing and communicatively connected to the multi-channel sensor array, the data processor being configured to perform real-time calibration of the data acquired by the dielectric constant sensor using a preset temperature compensation algorithm, and to perform pollutant concentration inversion on the data acquired by the miniature XRF (X-ray Fluorescence) probe using a preset random forest regression model; a memory communicatively connected to the data processor, the memory being used to store the calibrated and inverted data; and a communication interface for transmitting the stored data to an external system.

[0063] Specifically, the electrochemical sensor uses a solid-state pH / EC composite electrode.

[0064] Specifically, the heavy metal content data should include at least in-situ quantitative data for lead, cadmium, and arsenic.

[0065] Specifically, the real-time calibration of sensor data is based on a temperature compensation algorithm and pollutant concentration inversion. The specific formula for the temperature compensation algorithm is as follows: Formula (1); in: This represents the calibrated true volumetric moisture content. The volumetric water content as originally measured by the sensor; The temperature compensation coefficient is an empirical constant determined by linear regression after the sensor was calibrated in target soil using multiple temperature gradients. Its unit is °C. -1 ; T represents temperature.

[0066] Specifically, pollutant concentration inversion is calculated using a pre-built Random Forest Regression (RF) model.

[0067] Specifically, the memory is divided into a cache area, a main memory area, and a read-only firmware area.

[0068] Specifically, the cache area is used to temporarily store data; the main storage area is used to store structured data; and the read-only firmware area is used to store device driver data.

[0069] Specifically, the communication interface is also used to provide file system support to ensure the integrity of data exported offline.

[0070] Example 4 This embodiment provides a specific application example of a plant disease and pest detection-diagnosis-control system, based on any of the above embodiments.

[0071] In this embodiment, the deployment scenario for the plant disease and pest detection-diagnosis-control system is a smart plant protection system in a park in Shenzhen. Details are as follows:

[0072] Background: A park in Shenzhen covers approximately 80 hectares, with vegetation mainly consisting of Murraya paniculata, camphor trees, and broadleaf privet. The park is adjacent to a main urban road, and historical soil monitoring shows that some areas have road surface and vegetation contamination. The management hopes to establish an intelligent system to achieve early warning and precise, green pest control.

[0073] Specific deployment: Intelligent monitoring and data acquisition module: Image monitoring: Five multispectral cameras (including visible light and near-infrared channels) with gimbals are deployed at the highest point in the park for daily scheduled patrols and photography; a high-precision infrared thermal imaging scan of the entire park is conducted weekly using drones.

[0074] Soil and root monitoring: Twenty in-situ monitoring points were set up within the park according to functional zoning (near road area, core area, and waterfront area). Each point was equipped with a multi-channel sensor array; the multi-channel sensor array included at least:

[0075] FDR dielectric constant sensor (for measuring volumetric water content).

[0076] Solid-state composite ion-selective electrode (for measuring pH and nitrate).

[0077] Miniature silicon drift detector XRF probe (for in-situ semi-quantitative analysis of lead, cadmium, and arsenic in soil, with an excitation voltage of 40kV).

[0078] Meteorological data monitoring unit: A small weather station was set up in the center of the park to collect data on temperature, humidity, light, wind, and precipitation.

[0079] Data processing and communication module: Multiple sensors converge to the park management office's data processing and communication module via LoRa WAN for data preprocessing (timestamp alignment, format standardization, and outlier filtering), and then transmit the encrypted data to the cloud platform via a 5G private network.

[0080] Pest and disease diagnosis and prediction module: Deployed in the cloud, and includes at least one image recognition sub-model based on LSTM and XGBoost algorithms and performing deep learning image recognition, a health association analysis model, and a pest and disease prediction sub-model.

[0081] The intelligent decision-making and control module generates decision commands and sends them via API to the park's intelligent water and fertilizer integrated machine, precision spraying drone, and automatic release device for natural enemy insects (ladybugs) deployed on tree trunks.

[0082] Detailed workflow and data examples for the core modules: 1. Multi-source data fusion and AI diagnostics After receiving a data packet from a monitoring point (number P05, near the main road) on a certain day, the platform initiates the diagnostic process: Input data: Image data: Multispectral camera data shows that among the three camphor trees in area P05, one of them has a canopy NDVI (Normalized Difference Vegetation Index) of 0.68 (normal should be >0.75), and infrared images show that its canopy temperature is 1.5℃ higher than the surrounding area.

[0083] Soil data: XRF probe inversion showed that the soil had a lead (Pb) content of 320 mg / kg (3 times higher than the background value) and a cadmium (Cd) content of 1.2 mg / kg; the soil moisture content was 18% and the pH value was 7.1.

[0084] Meteorological data: The average humidity over the past week was 82%, and the average daily temperature was 25-28℃.

[0085] AI analysis models for collaborative diagnosis: Image recognition sub-model (based on random forest regression): Analyzing visible light images, multiple orange-red lesions were identified on the middle and lower leaves of the target tree, initially classified as "suspected leaf spot disease" with a confidence level of 75%. The image recognition sub-model was pre-trained to study the relationship between heavy metal stress and plant physiology. The health association analysis model, after receiving soil Pb and Cd data transmitted from a multi-sensor array, output a "heavy metal stress index" of 0.65 (moderate stress), and correlated it with the following: Under this stress level, the stomatal conductance of plant leaves may decrease, and the expected physiological resistance to the disease is reduced by approximately 25%.

[0086] Disease prediction sub-model (based on LSTM network): Combining the current high humidity and temperature, historical disease data for the same period, and the above-mentioned "weakened plant resistance" factor, it predicts that "the probability of a large-scale outbreak of camphor leaf spot disease in the P05 area is 80% within the next 7 days" and generates an orange warning.

[0087] Diagnostic results output: {Location: P05, Target: Camphor_001, Diagnosis: Camphor leaf spot disease (early stage), Risk level: Orange, Related factors: Moderate stress from soil heavy metals (Pb / Cd) leading to decreased plant resistance}.

[0088] 2. Intelligent decision-making and strategy optimization The intelligent decision-making and control module receives the above diagnostic results and initiates a multi-objective optimization decision tree: Decision constraints: The optimization objectives set by the park management are: ① Prevention and control effect > 80%; ② Minimize the amount of chemical pesticides used (high priority); ③ Controllable cost per operation.

[0089] Strategy generation and optimization: Rule base matching: Since the diagnosis includes "heavy metal stress", the system first excludes strong chemical pesticide solutions that may increase the metabolic burden on plants.

[0090] Multi-strategy simulation: Option A (Biological Control): Spraying with 80% mancozeb wettable powder is recommended. Simulated results: Environmentally friendly, but the expected control efficacy on plants with decreased resistance is approximately 70%, which does not meet the target.

[0091] Option B (Integrated Ecological and Agronomic Regulation): A combined strategy of "seaweed extract (inducing resistance) + potassium silicate solution (strengthening cell walls) + root application of dicalcium hydrogen phosphate (passivating heavy metals in the soil)" is recommended. Simulated results: Expected control efficacy of 85%, while gradually improving the rhizosphere environment, meeting all long-term goals.

[0092] Decision output: The system selects solution B as the optimal solution. Two executable instructions are generated:

[0093] Command 1 (Above-ground processing): {Equipment: Precision spraying drone, Target coordinates: [x1,y1], Agent: Seaweed extract (0.1%) + Potassium silicate (2%), Dosage: 15L / acre, Execution time: 10:00 the next day}.

[0094] Instruction 2 (Soil Treatment): {Equipment: Intelligent water and fertilizer integrated machine_zone 5, Operation: Inject dicalcium hydrogen phosphate solution, Dosage: 50kg / mu, Depth: 20cm}.

[0095] 3. Automated Execution and Feedback Learning Execution: The automated execution interface module converts instructions into device control codes. The following day, the drone completes foliar spraying according to the preset path; the integrated water and fertilizer machine starts up, injecting the pesticide into the P05 area irrigation system.

[0096] Feedback and Learning: A week later, the system re-analyzed the image data, and the NDVI index rose to 0.72. No new lesions were found, and the treatment was deemed "effective".

[0097] The entire data chain of this incident, from "multi-source data → diagnosis → decision → execution → result," was encrypted and stored, tagged with "successful case of leaf spot disease prevention and control under heavy metal stress," and used as data for AI analysis model analysis, learning, and training.

[0098] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for detecting, diagnosing, and controlling plant diseases and pests, characterized in that, Includes the following steps: S1. Acquire multi-source monitoring data on plant growth and pests and diseases, including plant image data, soil physicochemical index data, plant physiological index data, and environmental meteorological data; S2. Based on the multi-source monitoring data, the plant health status is diagnosed and the risk of pests and diseases is predicted through an AI analysis model, and diagnostic results and early warning information are generated. S3. Based on the diagnostic results and early warning information, generate and recommend at least one prevention and control strategy; S4. Through the intelligent decision-making and control module, the recommended prevention and control strategies are optimized, and executable control instructions are generated.

2. The plant disease and pest detection-diagnosis-control method according to claim 1, characterized in that, The steps for acquiring multi-source monitoring data include: Plant image data and temperature data acquired through the plant image monitoring unit; The soil physicochemical index data obtained through the soil and root monitoring unit includes one or more of the following: pH value, water content, electrical conductivity, heavy metal content, root growth status, root oxygen density, and total carbon content around the roots. Plant physiological data obtained through the plant index measurement module, including one or more of chlorophyll content and plant growth rate; Meteorological data obtained or retrieved through meteorological data monitoring units.

3. The plant disease and pest detection-diagnosis-control method according to claim 2, characterized in that, The soil and root monitoring unit includes a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe. The soil volumetric water content data collected by the dielectric constant sensor is calibrated in real time using a temperature compensation algorithm.

4. The plant disease and pest detection-diagnosis-control method according to claim 1, characterized in that, The prevention and control strategies generated by the intelligent decision-making and control module include at least one of the following: physical control strategy, biological pesticide application strategy, intelligent chemical pesticide application strategy, ecological agronomic prevention and control strategy, or strategy of requesting human intervention.

5. The plant disease and pest detection-diagnosis-control method according to claim 1, characterized in that, The AI ​​analysis model includes: An image recognition sub-model is used to classify and locate plant images in order to identify the types of pests and diseases; The disease prediction sub-model is used to predict the risk of disease outbreaks based on environmental meteorological data and historical data, using long short-term memory networks or extreme gradient boosting algorithms. And / or an insect infestation prediction sub-model, used to predict insect population density trends based on insect images and trajectory data.

6. A plant disease and pest detection-diagnosis-control system for implementing the method of any one of claims 1 to 5, characterized in that, include: The intelligent monitoring and data acquisition module is used to acquire multi-source monitoring data of plants through multiple sensors and image acquisition devices deployed in the monitoring area; The data processing and communication module is communicatively connected to the intelligent monitoring and data acquisition module, and is used to receive and preprocess the multi-source monitoring data, and transmit the processed data through the communication network. The pest and disease diagnosis and prediction module includes at least one AI analysis model. The pest and disease diagnosis and prediction module is communicatively connected to the data processing and communication module. Based on the data transmitted by the data processing and communication module, the pest and disease diagnosis and prediction module diagnoses the plant health status through the AI ​​analysis model and outputs diagnostic results and early warning information. The intelligent decision-making and control module is communicatively connected to the pest and disease diagnosis and prediction module. Based on the received diagnosis results and early warning information, the intelligent decision-making and control module generates at least one targeted prevention and control strategy. An automated execution interface module is communicatively connected to the intelligent decision-making and control module. The automated execution interface module drives one or more external execution devices to perform corresponding physical control operations according to the prevention and control strategy.

7. The plant disease and pest detection-diagnosis-control system according to claim 6, characterized in that, The intelligent monitoring and data acquisition module includes: The plant image monitoring unit is used to collect visible light and infrared thermal images of plants via drones or fixed cameras. The soil and root monitoring unit is used to monitor soil pH, moisture content, electrical conductivity, heavy metal content, root growth density, and oxygen density through in-situ probes. Plant physiological indicator monitoring unit, used to monitor the chlorophyll content and growth rate of plants; The meteorological data monitoring unit is used to acquire temperature, humidity, light, and precipitation data for the monitored area.

8. The plant disease and pest detection-diagnosis-control system according to claim 6, characterized in that, The external execution devices driven by the automated execution interface module include intelligent spraying systems, drones, integrated water and fertilizer equipment, predator release devices, or warning sign triggering devices.

9. A data acquisition device for monitoring plant diseases and pests, characterized in that, include: case; A multi-channel sensor array, integrated on the housing or connected to the housing via a cable, includes a dielectric constant sensor, an electrochemical sensor, and a miniature XRF probe, for in-situ acquisition of soil volumetric water content, dielectric constant, pH value, electrical conductivity, and heavy metal content data; A data processor, built into the housing and communicatively connected to the multi-channel sensor array, is configured to perform real-time calibration of the data collected by the dielectric constant sensor using a preset temperature compensation algorithm, and to perform pollutant concentration inversion on the data collected by the miniature XRF probe using a preset random forest regression model. A memory, communicatively connected to the data processor, is used to store the calibrated and inverted data; A communication interface is used to transfer stored data to an external system.

Images