Intelligent forestry insect situation forecasting system and device driven by image recognition and control decision

Abstract

The invention discloses an intelligent forestry insect situation forecasting system and device driven by image recognition and control decision, and belongs to the technical field of forestry informatization and pest control. The system comprises a sensing layer, a transmission layer, a processing layer, a decision-making layer and an execution layer. The sensing layer collects insect condition images and environment data through a special device; the processing layer automatically identifies pests by using an improved deep learning algorithm and predicts the pest situation trend by fusing multi-source data; a decision-making layer intelligently generates a prevention and treatment scheme based on an analysis result, and automatically dispatches execution equipment such as an unmanned aerial vehicle and a sprayer; and the execution layer completes prevention and control operation and feeds back an effect. The matched device integrates insect trapping, processing and shooting, and has the functions of rain and insect separation, intelligent control and the like. According to the method, full-process automation and intelligent closed-loop management of forestry insect conditions from monitoring, early warning to prevention and control execution is realized, the forecasting accuracy and prevention and control timeliness are remarkably improved, and the method is suitable for forestry application scenes of different scales.

Description

Technical Field

[0001] This invention belongs to the field of forest pest monitoring and control technology, specifically involving an intelligent forest pest monitoring and forecasting system and device driven by image recognition and control decision-making. Background Technology

[0002] As a core component of the ecosystem, the health of forestry directly affects ecological balance and economic development. However, outbreaks of forest pests often deal devastating damage to forestry resources. Therefore, timely monitoring and effective control of pests are crucial to ensuring the healthy development of forestry.

[0003] Current forestry pest monitoring technologies mainly rely on manual patrols and traditional monitoring equipment. Manual patrols are inefficient, have a narrow coverage area, and are greatly affected by human experience, making it difficult to achieve large-scale, all-weather pest monitoring. They are also prone to problems such as missed reports, false reports, and delayed monitoring. While traditional monitoring and forecasting equipment can achieve some automated trapping and data acquisition functions, such as attracting pests with insect-attracting light sources, processing insect bodies with far-infrared technology, and acquiring images, it still has many shortcomings: First, the image recognition accuracy is low, relying mostly on simple image comparison, making it difficult to accurately distinguish pest species with similar morphologies, and it has poor adaptability to insect posture and degree of damage; Second, the data processing capability is weak, unable to achieve deep integration and analysis of pest data and environmental data, making it difficult to accurately predict pest outbreak trends; Third, there is a lack of an effective linkage mechanism for prevention and control decisions, resulting in a disconnect between monitoring and forecasting data and prevention and control execution, and an inability to automatically generate targeted prevention and control plans based on real-time pest dynamics and drive equipment operation; Fourth, the equipment compatibility is poor, with different types of monitoring and control equipment finding it difficult to interconnect, forming data silos and affecting the collaborative efficiency of monitoring and control.

[0004] For example, while existing insect pest monitoring instruments possess functions such as high-definition cameras and wireless transmission, they can only capture, image, and upload data of insects. They cannot perform accurate intelligent identification of the uploaded insect images, nor can they combine environmental data such as temperature, humidity, light, and rainfall to analyze insect pest trends, let alone directly drive control equipment to carry out precise control operations. Furthermore, most existing control systems operate independently, requiring manual development of control plans based on monitoring data and manual scheduling of equipment, resulting in delayed control responses and difficulty in effectively controlling the spread of insect infestations.

[0005] Therefore, developing an intelligent forestry pest monitoring and forecasting system and device driven by image recognition and prevention and control decisions, and realizing full-process automation and intelligence from pest monitoring, identification, trend prediction to prevention and control execution, has become an urgent technical problem to be solved in the field of forestry pest monitoring and forecasting. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the existing defects and provide an intelligent forest pest monitoring and forecasting system and device driven by image recognition and prevention and control decisions, so as to solve the problems of low accuracy of forest pest monitoring and forecasting, weak data processing capabilities, poor prevention and control linkage and low level of intelligence mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides an intelligent forest pest monitoring and forecasting system driven by image recognition and prevention and control decisions, comprising a perception layer, a transmission layer, a processing layer, a decision layer and an execution layer connected in sequence;

[0009] The sensing layer is used to collect raw data of the target forestry area, including at least one insect pest monitoring device, an environmental sensor group, and an equipment status monitoring module. The insect pest monitoring device includes an insect-attracting module for attracting pests, an insect body processing module for killing and drying pests, and an image acquisition module for acquiring high-definition images of the processed insects. The environmental sensor group is used to collect at least one parameter among temperature, humidity, light intensity, rainfall, and wind speed. The equipment status monitoring module is used to monitor the operating status of each device in the sensing layer and the execution layer.

[0010] The transmission layer is used to realize secure data transmission between the perception layer, processing layer, decision layer and execution layer, and supports 4G, 5G wireless communication and / or Ethernet wired communication.

[0011] The processing layer is used for intelligent analysis of the data collected by the perception layer, including a data preprocessing module, an image recognition module, and a data fusion analysis module. The data preprocessing module is used to standardize and clean the insect infestation images and environmental parameter data. The image recognition module constructs an insect identification model based on an improved deep learning target detection algorithm to automatically identify the types and numbers of insects in the images. The data fusion analysis module combines the identified insect infestation data with environmental parameters and historical data, and uses a prediction algorithm to predict the trend of insect infestation.

[0012] The decision-making layer is used to generate executable prevention and control decisions based on the analysis results of the processing layer. It includes a prevention and control plan generation module and an equipment scheduling module. The prevention and control plan generation module matches or generates personalized prevention and control plans containing prevention and control methods, pesticides, time and scope from a preset plan library according to the pest species, quantity, predicted trend and regional forest stand characteristics. The equipment scheduling module is used to parse the prevention and control plan into specific operation instructions and issue them to the designated execution layer equipment.

[0013] The execution layer is used to receive and execute the operation instructions issued by the decision-making layer, including at least one of the following: drone control equipment, ground spraying equipment, predator release equipment, and physical trapping equipment. After execution, the execution layer equipment feeds back operation status data.

[0014] Furthermore, the image recognition module employs an improved deep learning target detection algorithm, namely the YOLOv8 algorithm with an attention mechanism, which is trained through transfer learning and data augmentation techniques to enhance the robustness of recognition under different postures, incomplete insect bodies, and complex lighting conditions.

[0015] Furthermore, the prediction algorithm used in the data fusion analysis module is a time series analysis algorithm or a neural network algorithm, which is used to fuse insect infestation time series data, environmental factors and UAV remote sensing vegetation index data to make multi-dimensional trend predictions.

[0016] Furthermore, the decision-making layer also includes an early warning module, which generates early warning information of different levels and pushes it to management personnel based on the predicted insect infestation trend and preset thresholds.

[0017] Furthermore, the system also includes a cloud platform module for centralized storage of system-wide data and for providing data visualization, historical query, statistical analysis, and remote management interfaces.

[0018] Furthermore, the decision-making layer also includes a decision optimization module, which is used to evaluate and dynamically optimize the effectiveness of the control schemes in the control scheme library based on the pest feedback data after the control operation.

[0019] Secondly, the present invention provides an insect pest monitoring device for the aforementioned system, comprising a cabinet, and integrated within the cabinet:

[0020] The insect-attracting module includes ultraviolet lamps of a specific wavelength and impact screens arranged around it, which are used to efficiently attract pests and make them fall into the processing channel;

[0021] The insect processing module includes at least one far-infrared processing chamber, a rain-insect separation device, and a conveying mechanism; the far-infrared processing chamber is used to kill the insects within a set time and dry them to a state suitable for identification; the rain-insect separation device is used to automatically separate rainwater from the insects; and the conveying mechanism is used to transport the processed insects to the shooting station in an orderly manner.

[0022] The image acquisition module includes a high-definition camera and a supplementary lighting unit, used to acquire high-definition feature images of the insect.

[0023] The control module is used to coordinate and control the automatic operation of the insect-attracting module, the insect treatment module, and the image acquisition module, and to collect data from the integrated sensor.

[0024] The power supply module supports both mains power and solar power modes, and features low-power standby and safety protection functions.

[0025] The data transmission module is used to interact with the processing layer or cloud platform of the system to exchange data and control commands.

[0026] Furthermore, the device also includes a human-machine interface for locally displaying device status and environmental parameters, and supporting manual control and parameter settings.

[0027] Furthermore, the device also integrates a positioning module for obtaining the device's precise geographic coordinates.

[0028] Furthermore, the rain-insect separation device adopts an inclined filter structure, allowing rainwater to be discharged through the filter while the insects are intercepted and fall into the treatment chamber.

[0029] Compared with existing technologies, this invention provides an intelligent forestry pest monitoring and forecasting system and device driven by image recognition and prevention and control decisions, which has the following beneficial effects:

[0030] 1. This invention constructs a pest identification model, introduces an attention mechanism to enhance the ability to extract key features, and combines transfer learning and data augmentation techniques to significantly improve the identification accuracy of pests with different postures, degrees of incompleteness, and lighting conditions, thus solving the problem of low pest identification accuracy in existing technologies.

[0031] 2. This invention constructs an intelligent architecture for the entire process of "sensing-transmission-processing-decision-execution", realizing automated collaboration of pest monitoring, identification, trend prediction, control decision generation and equipment scheduling execution. It can complete the entire process management of pest monitoring and control without manual intervention, thereby improving the efficiency of monitoring and control.

[0032] 3. This invention integrates insect pest data and environmental parameter data through a data fusion analysis module, and combines it with a historical insect pest database to predict trends, making insect pest forecasting more forward-looking and providing accurate data support for the formulation of prevention and control decisions;

[0033] 4. The decision-making layer of this invention can directly drive the execution layer equipment to carry out targeted prevention and control operations, realize the seamless connection between monitoring and forecasting data and prevention and control execution, effectively shorten the prevention and control response time, and improve the pest control effect;

[0034] 5. This invention supports multiple transmission and power supply methods, is compatible with different types of monitoring and control equipment, and can be flexibly expanded according to the actual needs of forestry areas, making it widely applicable;

[0035] 6. The device of this invention has a compact structure and is easy to install. It has functions such as rain-insect separation and light-controlled automatic operation. It can adapt to complex forestry field environments, and the cloud platform module supports remote access and management, making it convenient for users to keep track of insect dynamics and equipment operation status in real time. Attached Figure Description

[0036] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention, but do not constitute a limitation thereof. In the drawings:

[0037] Figure 1 This is a schematic diagram of the insect monitoring cabinet structure proposed in this invention;

[0038] Figure 2 This is a side view of the insect monitoring cabinet proposed in this invention;

[0039] Figure 3 This is the insect infestation control diagram proposed in this invention;

[0040] Figure 4 This is a system topology diagram proposed in this invention;

[0041] Figure 5 This is a diagram of the main interface of the touch display screen proposed in this invention;

[0042] Figure 6 This is a diagram of the system settings interface proposed in this invention;

[0043] Figure 7 This is a diagram of the cloud platform login interface proposed in this invention;

[0044] Figure 8 This is a diagram of the cloud platform command screen monitoring and early warning interface proposed in this invention;

[0045] Figure 9 This is a diagram of the task scheduling interface of the cloud platform command screen proposed in this invention;

[0046] Figure 10 This is a diagram of the cloud platform monitoring device management interface proposed in this invention;

[0047] Figure 11 This is a diagram of the cloud platform monitoring data management interface proposed in this invention;

[0048] Figure 12 This is a diagram of the cloud platform pest monitoring report interface proposed in this invention;

[0049] Figure 13 This is a diagram of the cloud platform monitoring and early warning information management interface proposed in this invention;

[0050] Figure 14 This is a diagram of the cloud platform data analysis interface proposed in this invention;

[0051] Figure 15 This is a diagram of the cloud platform sample plot management interface proposed in this invention;

[0052] Figure 16 This is a diagram of the cloud platform user management interface proposed in this invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Please see Figures 1-16 The present invention provides a technical solution: The present invention provides an intelligent forest pest monitoring and forecasting system driven by image recognition and prevention and control decisions, comprising a perception layer, a transmission layer, a processing layer, a decision layer and an execution layer connected in sequence;

[0055] The perception layer is used to collect raw data of the target forestry area, including at least one insect monitoring device, an environmental sensor group, and an equipment status monitoring module. The insect monitoring device includes an insect-attracting module for attracting pests, an insect body processing module for killing and drying pests, and an image acquisition module for acquiring high-resolution images of the processed insects. The environmental sensor group is used to collect at least one parameter among temperature, humidity, light intensity, rainfall, and wind speed. The equipment status monitoring module is used to monitor the operating status of each device in the perception layer and the execution layer.

[0056] The transport layer is used to enable secure data transmission between the perception layer, processing layer, decision layer and execution layer, and supports 4G and 5G wireless communication and / or Ethernet wired communication.

[0057] The processing layer is used for intelligent analysis of the data collected by the perception layer, including a data preprocessing module, an image recognition module, and a data fusion analysis module. The data preprocessing module is used to standardize and clean the insect infestation images and environmental parameter data. The image recognition module builds an insect identification model based on an improved deep learning target detection algorithm to automatically identify the types and numbers of insects in the images. The data fusion analysis module combines the identified insect infestation data with environmental parameters and historical data, and uses prediction algorithms to predict the trend of insect infestation.

[0058] The decision-making layer is used to generate executable prevention and control decisions based on the analysis results of the processing layer. It includes a prevention and control plan generation module and an equipment scheduling module. The prevention and control plan generation module matches or generates personalized prevention and control plans containing prevention and control methods, pesticides, time and scope from a preset plan library based on the pest species, quantity, predicted trend and regional forest stand characteristics. The equipment scheduling module is used to parse the prevention and control plan into specific operation instructions and issue them to the designated execution layer equipment.

[0059] The execution layer is used to receive and execute operational instructions issued by the decision-making layer, including at least one of drone-based pest control equipment, ground spraying equipment, predator release equipment, and physical trapping equipment. After execution, the execution layer equipment feeds back operational status data.

[0060] In this invention, preferably, the improved deep learning target detection algorithm used in the image recognition module is the YOLOv8 algorithm with an attention mechanism, which is trained through transfer learning and data augmentation techniques to improve the robustness of recognition under different postures, incomplete insect bodies, and complex lighting conditions.

[0061] In this invention, preferably, the prediction algorithm used in the data fusion analysis module is a time series analysis algorithm or a neural network algorithm, which is used to fuse insect infestation time series data, environmental factors and UAV remote sensing vegetation index data to make multi-dimensional trend predictions.

[0062] In this invention, preferably, the decision-making layer also includes an early warning module, which is used to generate early warning information of different levels and push it to the management personnel based on the predicted insect infestation trend and preset threshold.

[0063] In this invention, preferably, the system also includes a cloud platform module for centralized storage of system-wide data and for providing data visualization, historical query, statistical analysis, and remote management interfaces.

[0064] In this invention, preferably, the decision-making layer further includes a decision optimization module, which is used to evaluate and dynamically optimize the effectiveness of the control schemes in the control scheme database based on the pest feedback data after the control operation.

[0065] This invention provides an insect pest monitoring device for the aforementioned system, comprising a cabinet, and integrated within the cabinet:

[0066] The insect-attracting module includes ultraviolet lamps of a specific wavelength and impact screens arranged around it, which are used to efficiently attract pests and make them fall into the processing channel;

[0067] The insect processing module includes at least one far-infrared processing chamber, a rain-insect separation device, and a conveying mechanism; the far-infrared processing chamber is used to kill the insects within a set time and dry them to a state suitable for identification; the rain-insect separation device is used to automatically separate rainwater from the insects; and the conveying mechanism is used to transport the processed insects to the shooting station in an orderly manner.

[0068] The image acquisition module includes a high-definition camera and a supplementary lighting unit, used to acquire high-definition feature images of the insect.

[0069] The control module is used to coordinate and control the automatic operation of the insect-attracting module, the insect treatment module, and the image acquisition module, and to collect data from the integrated sensor.

[0070] The power supply module supports both mains power and solar power modes, and features low-power standby and safety protection functions.

[0071] The data transmission module is used to interact with the system's processing layer or cloud platform to exchange data and control commands.

[0072] In this invention, preferably, the device further includes a human-machine interface for locally displaying device status and environmental parameters, and supporting manual control and parameter settings.

[0073] In this invention, preferably, the device also integrates a positioning module for obtaining the device's precise geographical coordinates.

[0074] In this invention, preferably, the rain-insect separation device adopts an inclined filter structure, so that rainwater can be discharged through the filter while the insects are intercepted and fall into the treatment chamber.

[0075] See Figures 1-2 The pest monitoring cabinet utilizes modern optical, electronic, and CNC technologies to achieve automated far-infrared processing of insects, conveyor belt transportation, and automatic operation of the entire system. Without human supervision, it can automatically complete tasks such as attracting, killing, dispersing, photographing, transporting, collecting, and draining pests. Then, using wireless and IoT technologies, it uploads real-time environmental weather and pest information to a designated agricultural cloud platform for analysis and prediction of pest occurrence and development, providing services for modern agriculture and meeting the needs of pest forecasting, prediction, and specimen collection.

[0076] The insect monitoring cabinet features two-layer far-infrared insect treatment chambers with a mortality rate of no less than 98% and an insect removal rate of no less than 95%. It utilizes rain-insect separation technology to automatically separate rainwater from insects. Light control technology ensures automatic activation at night and automatic shutdown during the day, preventing changes in operating status due to sudden strong light. A 5-megapixel high-definition camera (optional 8-megapixel, 12-megapixel, and 20-megapixel resolutions) clearly identifies each insect. Built-in positioning allows users to view equipment data on a platform map. The insect-attracting lamp, insect-killing chamber, drying chamber, and camera can be remotely or manually controlled. The impact screens are angled at 120 degrees, with single-screen dimensions of 595±2mm (length), 213±2mm (width), and 5mm (thickness). The touchscreen displays the current operating mode, communication status, real-time temperature of the insect-killing and drying chambers, current light intensity, rainfall status, and the current operating status of each component.

[0077] See Figure 5 This interface displays illuminance, rainfall status, insecticidal chamber temperature, drying chamber temperature, equipment operating mode, and equipment address code. It also shows the equipment's operating status, including the insect-attracting lamp, insect rain baffle, drying baffle, insect transfer device, vibration device, supplemental lighting, insecticidal control, drying control, and insecticidal baffle. The equipment can also be switched to manual mode for individual control of each component.

[0078] See Figure 6 The system settings interface allows configuration of parameters such as start time, working duration, insect attraction duration, insect-attracting lamp on-time duration, post-rain delay on-time, insecticidal drying chamber temperature, device target address, and port. Target parameter settings: Target port: Our environmental monitoring cloud platform's real-time data listening port is 8045, and the image receiving port is 8077. If the host uploads data to our cloud platform, the target port should be set to 8045, and the image upload port should be set to 8077. Target address: The IP address or domain name of the computer or server where the monitoring platform is located. If the device uploads data to our cloud platform...

[0079] Please see Figure 7 The cloud platform login link can be accessed by entering the assigned username and password. Please refer to [link / reference]. Figure 8 Monitoring homepage: This page displays the device's location information. Please refer to [link / reference]. Figure 9 Real-time pest situation: View images of pests captured by the device, including the area where the device is located, device name, sampling time, and number of pests. (See also...) Figure 10 Insect Infestation Analysis: Analyzes pest information in images captured by the equipment. You can choose "automatic identification" or manual identification. (See also...) Figure 11 Pest Species: This section identifies each type of insect. It facilitates easy reference and recording, can be directly accessed in pest reports, or searched by insect name. See also... Figure 12 Statistical analysis: You can view the regional pest statistics and the pest quantity change trend of the equipment within a specified time period.

[0080] See Figures 13-14 Equipment monitoring: The device status can be viewed in real time, and the device's operating status can be switched by clicking the operating mode. When the device is in automatic mode, it will work automatically when the set program reaches the working time period. When the device is in manual mode, the device can be controlled by clicking on "Insect attractant lamp status", "Insect rain baffle", "Insect killing baffle", "Drying baffle", "Insect transfer device", "Vibration device", "Supplemental light", "Camera" and other controls.

[0081] See Figures 15-16 Equipment Management: Click "System Management," select "Equipment Management," and then click "Equipment Information" to modify "Equipment Name," "Equipment Latitude and Longitude," "Offline Judgment Time," and "Data Storage Interval," etc. You can also modify equipment parameters in automatic mode, such as "Start Time," "Working Duration," and "Insect Attraction Duration."

[0082] Example 1: Basic System for Small and Medium-Sized Forest Areas

[0083] This embodiment provides a basic system suitable for small and medium-sized forestry areas at the county and town levels.

[0084] 1. System Composition:

[0085] Sensing layer: Deploy 3 insect monitoring devices (5-megapixel cameras) and one set of environmental sensors (temperature, humidity, light intensity, and rainfall). The devices are arranged in a triangular pattern to cover the core monitoring area.

[0086] Transport layer: 4G DTU is used for wireless data transmission.

[0087] Processing layer: Deployed on edge servers. The image recognition module uses a lightweight YOLOv8-nano model, trained on five major local pests including pine caterpillars and fall webworms, achieving an accuracy rate >92%. The data fusion and analysis module uses the ARIMA algorithm, combining insect population data with environmental temperature and humidity to predict trends for the next 3-5 days.

[0088] Decision-making level: Deployed on the same server. The preset control plan library includes two categories: chemical control (for high insect population density) and physical / manual control (for low insect population density). Warning thresholds are set as follows: Blue (20-50 insects / day), Orange (50-100 insects / day), and Red (>100 insects / day). When the threshold is reached, the system notifies the administrator via SMS and platform message, and automatically generates a control recommendation form containing the work area, recommended pesticide, and dosage.

[0089] Execution layer: Equipped with 2 intelligent ground sprayers, which can receive operation instructions (including GIS boundaries of the work area and spraying parameters) wirelessly issued by the system.

[0090] 2. Supporting Equipment: The system uses a standard insect monitoring cabinet (approximately 717mm × 727mm × 1565mm), equipped with a 20W 365nm UV lamp, a 120° three-sided impact screen, a double-layered far-infrared processing chamber (operating temperature 85℃), a tilting filter-type insect separation device, a 5-megapixel camera, and LED supplementary lighting. The controller is an STM32 series MCU, equipped with a 7-inch touchscreen for local monitoring. Power supply is 220V AC mains.

[0091] 3. Workflow: At night, the device automatically turns on the insect-attracting lights. Pests are attracted, collide with, and fall into the treatment chamber, where they die and are dried. They are then transported by conveyor belt to the imaging area for photography. Images and sensor data are uploaded to the processing layer via 4G. If 65 pine caterpillars are detected per day, an orange alert is triggered. The decision-making layer matches the "ground spray chemical control" plan and automatically issues operation instructions to the sprayers in the designated areas. After administrator confirmation on the platform, the sprayers can execute the operation automatically or semi-automatically. Pest monitoring continues after the operation, forming a closed loop.

[0092] Example 2: Enhanced System for Large-Scale Forest Farms

[0093] This embodiment is an enhancement of Embodiment 1 and is applicable to large state-owned forest farms at the provincial and municipal levels.

[0094] 1. System Enhancement Points:

[0095] Enhanced perception layer: Add a drone remote sensing monitoring unit (equipped with a multispectral camera) to regularly patrol and acquire NDVI (vegetation index) data of the forest area to assist in macro-level insect infestation assessment.

[0096] Processing layer enhancements: The image recognition model has been upgraded to YOLOv8s, expanding the number of pest species it can identify to 15. The data fusion and analysis module has been upgraded to use an LSTM neural network, fusing pest data, six environmental data points (temperature, humidity, light, rain, wind, and air pressure), and NDVI data to predict the probability and extent of pest outbreaks in the next 7-10 days.

[0097] Enhanced decision-making capabilities: Integrated into a cloud-based SaaS platform. The prevention and control solution library has been expanded to include biological control (such as releasing natural enemy insects) and aerial control solutions. The equipment scheduling module possesses complex task orchestration capabilities, enabling the coordinated scheduling of multiple drones and ground equipment for multi-dimensional operations.

[0098] Enhanced execution layer: Equipped with a fleet of agricultural drones (with a payload of over 10L), large ground sprayers, and insect predator release devices.

[0099] 2. Enhanced Supporting Equipment: The camera has been upgraded to 12 megapixels, and a GPS / BeiDou dual-mode positioning module has been added. The power supply system has been upgraded to a hybrid power supply of "mains power + solar power" and is equipped with a high-capacity battery. The cabinet is made of stainless steel with an IP65 protection rating.

[0100] 3. Example of Intelligent Decision-Making Process: The system, through insect monitoring devices and drone remote sensing, comprehensively judges that the population density of longhorn beetles, the vector insects of pine wilt nematode disease, is increasing in a certain pine forest, and that NDVI shows a decline in the health of local trees. The data fusion model predicts that the next two weeks will be a high-risk period for spread. The decision-making layer automatically generates a comprehensive plan of "precise drone spraying of pesticides to block the spread + hanging traps in the forest". The scheduling module plans the drone operation grid route and issues instructions, while simultaneously pushing specific location navigation for hanging traps to the mobile terminals of management personnel. All operation progress and effect feedback are summarized in real time on the cloud platform screen.

[0101] Example 3: Lightweight System for Park Nurseries

[0102] This embodiment provides a low-cost, easy-to-deploy lightweight solution suitable for small areas such as urban parks and nurseries.

[0103] 1. Simplified system design:

[0104] Sensing layer: Only one lightweight insect monitoring and forecasting device is deployed.

[0105] Transmission and processing layer: Employs 4GCat.1 low-power transmission. Image recognition and data analysis functions are fully hosted on a public cloud AI service platform, eliminating the need for users to build their own servers.

[0106] Decision-making level: Extremely simplified. After identifying the insect infestation, the cloud service directly sends an early warning notification and simple prevention and control suggestions to the administrator's WeChat mini program (such as "15 adult Sophora japonica looper have been found. It is recommended to manually shake the branches to kill them tomorrow morning").

[0107] Execution layer: mainly relies on manual prevention and control; the system does not directly control large equipment.

[0108] 2. Lightweight supporting equipment: The cabinet size has been reduced (approximately 400mm × 350mm × 800mm) and is made of engineering plastic. The insect-attracting lamp power has been reduced to 10W, and the insect treatment chamber has been simplified to a single layer. An 8-megapixel fixed-focus camera is used. Power is supplied by a solar panel and a built-in battery, enabling completely off-grid operation.

[0109] 3. Workflow: The device operates automatically, uploading images of the insects captured daily to the cloud. After identification by cloud AI, if the number exceeds a preset threshold (e.g., 10 insects / day), an alarm message is immediately sent to the designated administrator. The administrator then organizes manual control measures based on the recommendations and provides feedback on the handling status via a mini-program.

[0110] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions, characterized in that, include: The sensing layer is used to collect insect infestation image data and environmental parameter data in forestry areas, and includes at least one insect infestation monitoring and forecasting device and an environmental sensor group. The transport layer is used to transmit the data collected by the sensing layer; The processing layer, used to analyze the received data, includes: The image recognition module automatically identifies the types and quantities of pests in insect images based on a deep learning target detection algorithm. The data fusion and analysis module is used to combine the identified insect infestation data with the environmental parameter data and output insect infestation trend prediction information; The decision layer, used to generate prevention and control decisions based on the identification results and trend prediction information output by the processing layer, includes: The prevention and control plan generation module is used to generate or match corresponding prevention and control plans based on pest species, quantity and trend prediction information. The equipment scheduling module is used to convert the prevention and control plan into executable equipment operation instructions; The execution layer includes at least one prevention and control device capable of receiving and executing operation instructions issued by the equipment scheduling module.

2. The intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions according to claim 1, characterized in that, The image recognition module uses the YOLOv8 algorithm, which incorporates an attention mechanism, as its deep learning object detection algorithm.

3. The intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions according to claim 1, characterized in that, The data fusion and analysis module is also used to access historical insect infestation databases and / or UAV remote sensing monitoring data, and to use time series analysis algorithms or neural network algorithms to predict trends.

4. The intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions according to claim 1, characterized in that, The decision-making layer also includes an early warning module, which generates early warning information based on the trend prediction information and preset thresholds, and sends it to the user terminal through a communication network.

5. The intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions according to claim 1, characterized in that, The system also includes a cloud platform module, which is used to centrally store the data of the perception layer, processing layer, decision-making layer and execution layer, and provide data visualization and remote management interfaces.

6. The intelligent forestry pest monitoring and forecasting system driven by image recognition and prevention and control decisions according to claim 1, characterized in that, The decision-making layer also includes a decision optimization module, which is used to evaluate and optimize the effectiveness of the control plan based on the pest feedback data after the execution layer has completed its work.

7. The pest monitoring device of the intelligent forestry pest monitoring system driven by image recognition and control decision-making according to any one of claims 1-6, characterized in that, Including the cabinet itself, and the components housed within it: Insect-attracting module, used to attract pests; The insect treatment module, connected to the insect-attracting module, is used to kill and dry the fallen pests, and includes a rain-insect separation device. An image acquisition module is located at the output end of the insect processing module and is used to acquire high-definition images of the processed insect. The control module is electrically connected to the insect-attracting module, the insect processing module, and the image acquisition module, and is used to control their coordinated operation. The power supply module provides power to all modules within the device. The data transmission module, connected to the control module, is used to upload the image data and device status data acquired by the image acquisition module.

8. The pest monitoring device of the intelligent forestry pest monitoring system driven by image recognition and prevention and control decisions according to claim 7, characterized in that, The insect processing module includes a far-infrared processing chamber and a conveying mechanism. The insect separation device is located at the inlet end of the far-infrared processing chamber and adopts an inclined filter structure.

9. The pest monitoring device of the intelligent forestry pest monitoring system driven by image recognition and prevention and control decisions according to claim 7, characterized in that, The device also includes a human-computer interaction interface and / or a positioning module.

10. The pest monitoring device of the intelligent forestry pest monitoring system driven by image recognition and prevention and control decisions according to claim 7, characterized in that, The power supply module supports both 220V AC mains power and solar power.

Images