Method for preventing and treating crop diseases based on precise irradiation of ultraviolet rays and cross-ridge operation equipment

Abstract

The present application relates to the field of agricultural intelligent equipment and plant protection technology, and discloses a method for preventing and treating crop diseases based on precise ultraviolet irradiation and cross-ridge operation equipment, comprising the following steps: a main control module drives the chassis to cruise autonomously, and a self-adaptive adjustment module adjusts the damping force in real time according to the attitude data; a light supplementing module dynamically adjusts the light to cooperate with the identification module to obtain disease information; the main control module calculates the triggering condition based on the running speed and the installation distance between the identification module and the ultraviolet irradiation array, controls the corresponding ultraviolet unit to carry out pinpoint irradiation on the disease target; and an operation traceability module generates encrypted tamper-proof data and uploads it. The present application realizes precise irradiation through odometer pulse synchronization and speed dose compensation, maintains the operation attitude stable under complex terrain through active suspension control, and realizes reliable traceability of operation data through chain encryption technology, solving the problems of lack of precision in traditional prevention and treatment, unstable equipment attitude and difficult data traceability.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment and plant protection technology, specifically to a method for preventing and controlling crop diseases based on precise ultraviolet irradiation and equipment for cross-ridge operations. Background Technology

[0002] Using short-wave ultraviolet (UV-C) radiation to disrupt the DNA structure of pathogens has become an important means of physical control of crop diseases. However, existing UV control equipment typically uses a constant power, full-coverage continuous irradiation mode, which cannot be differentiated according to the distribution of diseases, and lacks a coordinated control logic for travel speed and radiation dose, resulting in low energy utilization, susceptibility of healthy crops to radiation stress, and uneven control effects.

[0003] Furthermore, unstructured terrain in the field causes the operating chassis to bump and tilt. Existing equipment lacks active attitude adjustment capabilities, which not only blurs the images from visual sensors but also drastically alters the relative distance between the light source and the crop surface. Given the characteristic that ultraviolet radiation intensity decreases with distance, this distance fluctuation causes the actual radiation energy to deviate from the effective threshold, resulting in control failure or pesticide damage. Simultaneously, existing methods of recording operational data lack secure verification mechanisms, making data susceptible to tampering or loss, and failing to meet the modern agricultural regulatory requirements for reliable traceability throughout the entire disease control process. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and cross-ridge operation equipment for the precise control of crop diseases based on ultraviolet irradiation. This solves the problems in existing physical control technologies for crop diseases, such as the lack of precision in ultraviolet irradiation leading to energy waste and crop damage, the undulating terrain of the field furrows causing the operating platform to vibrate and thus reducing the accuracy of identification, and incomplete operation data recording leading to the inability to trace the source.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling crop diseases based on precise ultraviolet irradiation, comprising the following steps: After receiving the operation start command, the positioning module parses the current location information. The main control module controls the hub motor to drive the autonomous mobile chassis to autonomously cruise between crop rows. During the movement, the adaptive adjustment module adjusts the damping force of the shock absorption damping according to the real-time collected attitude data. The supplemental lighting module dynamically adjusts the luminous intensity according to the ambient light intensity, the recognition module continuously acquires crop images, and the main control module runs the detection model to output the pixel coordinates and disease level of the disease target. The main control module uses time synchronization logic to calculate the number of target trigger pulses when the diseased target enters the coverage area of ​​the UV-C lamp array, based on the real-time travel speed of the autonomous mobile chassis and the physical installation distance between the identification module and the UV-C lamp array. When the diseased target reaches the irradiation position, the main control module calculates the required target radiation dose according to the disease level and controls the ultraviolet emitting unit at the corresponding position in the UV-C lamp array to perform fixed-point irradiation. Throughout the entire operation, the operation traceability module synchronously collects operation data, uses encryption technology to generate tamper-proof data blocks, and uploads them via antenna.

[0006] Preferably, in the adjustment step of the supplementary lighting module, the main control module acquires ambient light intensity data, reads the imaging illumination intensity threshold, and calculates the difference between the imaging illumination intensity threshold and the current ambient light intensity; the main control module generates a pulse width modulation signal duty cycle based on the difference, drives the LED array of the supplementary lighting module to supplement light, and ensures that the recognition accuracy of the detection model is not affected by changes in natural illumination by maintaining the consistency of the imaging illumination environment.

[0007] Preferably, the fixed-point irradiation control adopts a synchronous control logic based on odometer pulses: the main control module obtains the current global pulse count value fed back by the hub motor, calculates the target trigger pulse number based on the longitudinal horizontal distance of the target in the vehicle coordinate system collected by the identification module, combined with the dynamic diameter of the moving wheel and the slip correction coefficient; the main control module establishes a task queue, pushes the task objects containing the target trigger pulse number, lateral coordinate and damage level into the queue, when the real-time pulse count reaches the target trigger pulse number, maps the corresponding position of the lamp column index in the UV-C lamp array according to the lateral coordinate, drives the ultraviolet emitting unit at the corresponding position to work, thereby realizing the accurate spatial mapping between the detection position and the irradiation position during dynamic movement.

[0008] Preferably, the calculation steps for the target radiation dose include: the main control module finding the minimum effective lethal energy density based on the disease level, calling the duty cycle calculation model, which uses the actual travel speed of the autonomous moving chassis as a speed compensation variable and the vertical distance between the crop surface and the LED beads as an attenuation correction variable for calculation; the main control module calculating the theoretical target duty cycle, and if the theoretical target duty cycle exceeds the physical upper limit, reducing the command speed of the autonomous moving chassis, using extended exposure time to compensate for radiation intensity, and ensuring that the accumulated radiation energy meets the needs of disease prevention and control.

[0009] Preferably, in the operation data processing step, the operation tracing module encapsulates the timestamp, location coordinates, disease image, and irradiation parameters into a current operation data frame; generates a random salt value; and uses a one-way hash function to generate an encrypted hash value for the current block. The input for calculating the encrypted hash value includes the content of the current operation data frame, the hash value of the previous data block, and the random salt value. A chained storage structure is constructed to prevent the operation record from being tampered with.

[0010] Preferably, the adaptive adjustment module adjusts the damping force of the shock absorber as follows: calculate the square root of the sum of the squares of the real-time detected roll angle and pitch angle as the comprehensive attitude error, use the PID algorithm to calculate the control quantity, and output an electrical signal to drive the damping force output by the shock absorber damping adjustment, thereby suppressing the vibration and tilt of the chassis on the uneven road surface and maintaining the stable working posture of the sensor and actuator.

[0011] A second aspect of this invention provides cross-row operation equipment for controlling crop diseases based on ultraviolet (UV) precision irradiation, used to execute the aforementioned method for controlling crop diseases based on UV precision irradiation. The equipment includes an autonomous mobile chassis. The main frame of the autonomous mobile chassis is an arched cross-row structure. A suspension system is symmetrically arranged on both sides of the bottom of the autonomous mobile chassis. A walking drive unit is located at the bottom of the suspension system. A positioning module and an antenna are located on the top of the autonomous mobile chassis. A power module is connected to the side of the autonomous mobile chassis. An integrated supplementary lighting and identification unit is located at the front end of the autonomous mobile chassis in the forward direction. The integrated supplementary lighting and identification unit integrates an identification module and a supplementary lighting module. The identification module is located next to the supplementary lighting module. An identification camera is located on the top of the autonomous mobile chassis. UV-C lamp arrays are arranged on the top and inner walls of both sides of the inner side of the arched frame of the autonomous mobile chassis. A main control module and an operation traceability module are located inside the autonomous mobile chassis.

[0012] Preferably, the suspension system includes a shock absorber and an adaptive adjustment module. One end of the shock absorber is mounted on the autonomous mobile chassis, and the other end is mounted on the adaptive adjustment module. The walking drive unit is located at the bottom of the adaptive adjustment module. The walking drive unit includes a wheel hub connector. One end of the wheel hub connector is located at the bottom of the adaptive adjustment module, and the other end of the wheel hub connector is equipped with a wheel hub motor. The output end of the wheel hub motor is fixedly connected to a moving wheel. This structure realizes the integrated arrangement of chassis support, shock absorption, and drive.

[0013] Preferably, the UV-C lamp array is composed of multiple independently controlled ultraviolet emitting units arranged in a matrix; a physically flexible light-blocking curtain is provided between the UV-C lamp array and the supplementary lighting module. The top of the physically flexible light-blocking curtain is fixed to the inner wall of the arched frame, and the bottom hangs down naturally, which is used to block the mutual interference between the supplementary lighting source and the ultraviolet light source, and reduce the scattering of ultraviolet rays into the external environment.

[0014] Preferably, the main control module is electrically connected or connected to the hub motor, adaptive adjustment module, identification module, supplementary lighting module, UV-C lamp array, positioning module and operation traceability module via a data bus; the power module includes a battery pack and a fuel generator, and the battery pack is electrically connected to each of the above electronic components.

[0015] This invention provides a method for controlling crop diseases based on precise ultraviolet irradiation and equipment for cross-ridge operation. It has the following beneficial effects: 1. This invention constructs a synchronous control logic based on odometer pulses to accurately map visual recognition coordinates to the coordinate system of the actuator, and eliminates time lag errors during mobile operations by combining the slip correction of the moving wheel; at the same time, the system can dynamically adjust the radiation dose according to the disease level and the travel speed, and automatically reduce the speed to extend the exposure time when the theoretical duty cycle is limited. Through the fixed-point precise application method, the system can ensure the lethal dose of the disease while avoiding the energy waste caused by traditional full-coverage irradiation, and reduce the stress damage caused by excessive ultraviolet radiation to the healthy tissues of crops.

[0016] 2. This invention features a suspension system with an adaptive adjustment module. By actively adjusting the output force of the shock-absorbing damping using real-time attitude data feedback, it suppresses the bumps and tilts of the autonomous mobile chassis when traveling on unstructured roads between rows. This ensures the imaging clarity and stability of the integrated front-end lighting and recognition unit, and also maintains a constant relative distance between the UV-C lamp array and the crop surface, avoiding missed detections or deviations in radiation attenuation model calculations caused by sudden changes in vehicle attitude.

[0017] 3. This invention introduces an operation traceability mechanism based on a cryptographic hash algorithm, which encapsulates irradiation parameters, geographical location, and disease images into a chain data structure and introduces random salt values ​​to prevent data collisions and tampering. Through data recording, it realizes credible evidence storage for the entire operation process, providing real and reliable original data support for the quality supervision of crop disease prevention and control and subsequent food safety traceability, and solving the problem that traditional agricultural operation data is easily lost or maliciously modified. Attached Figure Description

[0018] Figure 1 This is a three-dimensional structural diagram of the cross-ridge operation equipment of the present invention; Figure 2 This is a schematic diagram showing the location distribution of the ultraviolet precision irradiation unit of the present invention; Figure 3 This is a schematic diagram of the method for preventing and controlling crop diseases based on precise ultraviolet irradiation, as provided in an embodiment of the present invention.

[0019] The components include: 1. Autonomous mobile chassis; 2. Shock absorption and damping; 3. Adaptive adjustment module; 4. Wheel hub connector; 5. Wheel hub motor; 6. Moving wheels; 7. Recognition module; 8. Supplemental lighting module; 9. UV-C LED array; 10. Positioning module; 11. Antenna; 12. Power module; 13. Main control module; 14. Operation traceability module; and 15. Recognition camera. Detailed Implementation

[0020] The technical solutions in 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.

[0021] Please see the appendix Figures 1-2 This invention provides cross-ridge operation equipment for the prevention and control of crop diseases based on ultraviolet precise irradiation, including an autonomous mobile chassis 1, a power module 12, a positioning module 10, an identification module 7, a supplementary lighting module 8, a UV-C lamp array 9, a main control module 13, and an operation traceability module 14.

[0022] The main frame of the autonomous mobile chassis 1 adopts an arched, cross-ridge structure adapted to the height of the crop canopy. Walking drive units are symmetrically arranged on both sides of the bottom of the autonomous mobile chassis 1. These walking drive units employ an all-wheel independent drive mode, with each moving wheel 6 mechanically connected to a corresponding hub motor 5 via a hub connector 4. The moving wheels 6, as the walking components in contact with the ground, are selected according to the soil environmental characteristics of the working area. These soil environmental characteristics are measured using soil volumetric moisture content as a standard, specifically through measurements using devices such as frequency domain reflectance (FDR) humidity sensors. In dryland environments where the measured soil volumetric moisture content is below a preset threshold, pneumatic rubber tires are used for the moving wheels 6; in paddy field environments where the measured soil volumetric moisture content is above the preset threshold, blade-type paddy field wheels are selected for the moving wheels 6.

[0023] The suspension system of the autonomous mobile chassis 1 includes a shock absorber damper 2 and an adaptive adjustment module 3. The adaptive adjustment module 3 integrates an attitude sensor, which collects roll and pitch angle data of the autonomous mobile chassis 1 in real time during travel. The shock absorber damper 2 is a variable damping shock absorber, and its control terminal is electrically connected to the adaptive adjustment module 3. The adaptive adjustment module 3 adjusts the damping coefficient of the shock absorber damper 2 based on the feedback data from the attitude sensor to maintain the level attitude of the autonomous mobile chassis 1.

[0024] An integrated lighting and recognition unit is installed at the forward-facing front of the autonomous mobile chassis 1. This integrated unit comprises a recognition module 7 and a lighting module 8. The recognition module 7 includes a high-resolution RGB camera for acquiring image data of the crop surface. The lighting module 8 includes a low-power LED array for providing visible light auxiliary illumination with wavelengths from 450nm to 550nm in low-light environments. A recognition camera 15 is also installed on the top of the autonomous mobile chassis 1 to identify its direction of movement, enabling better planning of the movement route and identification of road conditions, thus making the movement of the autonomous mobile chassis 1 more precise.

[0025] An ultraviolet (UV) precision irradiation unit is installed on the inner side of the autonomous mobile chassis 1. This unit comprises a UV-C lamp array 9, which is distributed on the inner top and both inner walls of the arched frame, forming an enclosed irradiation field. A flexible physical light-blocking curtain is installed between the UV-C lamp array 9 and the supplementary lighting module 8 to block light interference from the supplementary lighting module 8 and prevent UV scattering. The UV-C lamp array 9 consists of multiple independently controlled UV emitting units, each with its own independent electromagnetic switch. The UV-C lamps in the UV-C lamp array 9 emit short-wave ultraviolet (Ultraviolet C) light.

[0026] A positioning module 10 and an antenna 11 are mounted on the top of the autonomous mobile chassis 1. The positioning module 10 is configured to receive signals from the Global Navigation Satellite System or analyze SLAM offline map data, and output the real-time latitude and longitude coordinates or relative position coordinates of the autonomous mobile chassis 1. The antenna 11 is connected to the communication unit for remote data transmission. The power module 12 is fixed to the autonomous mobile chassis 1 and includes a battery pack and a fuel generator. The fuel generator is electrically connected to the battery pack for real-time power replenishment, and the battery pack provides power to all electronic components of the machine.

[0027] The main control module 13 is installed in the electrical control box of the autonomous mobile chassis 1, and establishes bidirectional data communication connections with the hub motor 5, adaptive adjustment module 3, identification module 7, supplementary lighting module 8, UV-C lamp array 9, positioning module 10, and operation traceability module 14. The main control module 13 has a built-in edge computing unit for executing path planning, defect identification algorithm inference, and issuing logic control commands. The operation traceability module 14 is used to collect and encrypt key data during the operation process.

[0028] See attached document Figure 3 Based on the aforementioned cross-ridge operation equipment, this invention provides a method for controlling crop diseases using precise ultraviolet irradiation, the overall operation process of which is as follows: The operation begins at night or in low-light conditions. The positioning module 10 analyzes the current location information, and the main control module 13 generates a planned path based on the preset operation boundary, and controls the hub motor 5 to drive the autonomous mobile chassis 1 to autonomously cruise between crop rows. During the journey, the adaptive adjustment module 3 adjusts the vehicle's posture in real time to ensure the spatial stability of the identification module 7 and the UV-C lamp array 9 relative to the crop canopy.

[0029] The recognition module 7 continuously acquires crop images. The supplementary lighting module 8 dynamically adjusts the light intensity of the LED array based on the values ​​fed back from the ambient light intensity sensor, providing stable imaging lighting conditions for the recognition module 7. The main control module 13 receives image data and runs the embedded ELS-YOLO (Efficient Lightweight Small-object YOLO) lightweight model. ELS-YOLO is a lightweight target detection model based on the YOLOv11s framework, with the core design goal of solving the problems of low target detection accuracy, high model redundancy, and insufficient real-time performance in low-light environments (such as nighttime or weak light). The main control module 13 uses this model to detect disease areas within the field of view in real time and outputs the pixel coordinates, area size, and disease severity of the disease areas.

[0030] The main control module 13 uses a Kalman filter algorithm and time synchronization logic to calculate the time delay for the affected area to enter the coverage area of ​​the UV-C lamp array 9, based on the real-time travel speed of the autonomous mobile chassis 1 and the physical installation distance between the identification module 7 and the UV-C lamp array 9. When the affected area reaches the irradiation position, the main control module 13 calculates the required target radiation dose according to the severity of the disease and controls the corresponding electromagnetic switch to activate the UV-C lamp array 9 in the specific area for targeted irradiation.

[0031] Throughout the entire operation, the operation traceability module 14 synchronously collects operation time, geographical coordinates, disease severity data, real-time output power of the UV-C lamp array 9, and cumulative irradiation dose. The operation traceability module 14 uses blockchain hash encryption technology to encrypt the above core data, generating tamper-proof data blocks, and uploads them to the cloud management platform via antenna 11, realizing digital traceability and supervision of the operation process.

[0032] The autonomous mobile chassis 1, serving as the motion carrier for cross-ridge operation equipment, adopts an all-wheel independent drive power architecture. Four independent drive wheel sets are symmetrically distributed at the bottom of the autonomous mobile chassis 1, each powered by a hub motor 5. The rotor shaft of the hub motor 5 is directly and rigidly connected to the rim of the mobile wheel 6 via a flange-structured hub connector 4. This design eliminates the complex mechanical transmission shafts, gearboxes, and differential components found in traditional agricultural machinery, achieving a shorter and more efficient power transmission chain.

[0033] As a walking component that comes into direct contact with the soil, the mobile wheel 6 is physically adapted to the soil environmental characteristics of the work area to meet the passability requirements under different terrains. Specifically, the work environment is determined based on soil moisture parameters. When the measured soil moisture value of the work area is less than or equal to a preset moisture threshold (set to 60% in this embodiment), the mobile wheel 6 uses an inflatable rubber tire with a herringbone pattern. This type of tire utilizes the elastic deformation of the rubber material to increase the ground contact area, providing sufficient ground adhesion and buffering the high-frequency vibrations of hard dryland surfaces. When the measured soil moisture value of the work area is greater than the preset moisture threshold (i.e., exceeding 60%), it indicates that the work environment is a paddy field or an excessively wet plot. In this case, the mobile wheel 6 uses a blade-type paddy field wheel or a high-tooth steel wheel. The wheel rim has several radially arranged metal blades or protruding steel teeth evenly distributed on its circumferential surface. During the rotation of the wheel, the metal blades can penetrate into the saturated soft soil layer and use the shear reaction force generated by the soil being squeezed to form an effective driving torque, thereby preventing the chassis from slipping or sinking in muddy and humid environments.

[0034] The motion control of the autonomous mobile chassis 1 relies on the pre-set kinematic calculation logic within the main control module 13. Considering the characteristics of independent all-wheel drive, the main control module 13 employs differential steering to achieve the chassis's planar motion. Using the commonly used differential kinematic model in this field, the main control module 13 decomposes the overall target linear velocity and target angular velocity of the chassis issued by the path planning layer into independent speed commands for the left and right wheel sets. Specifically, the target linear velocity of the left wheel set is calculated as the algebraic difference between the chassis centerline velocity and the tangential velocity component generated by the steering angular velocity at half the wheel track, while the target linear velocity of the right wheel set is calculated as the algebraic sum of the two. Based on the effective rolling radius of the wheels, the main control module 13 converts these linear velocity components into the target rotational angular velocities of each hub motor 5.

[0035] To ensure accurate tracking of the target command at the actual travel speed, the hub motor 5 integrates position and speed feedback elements, such as Hall sensors or incremental photoelectric encoders, to collect real-time data on the actual rotational speed of the motor rotor. The main control module 13 constructs a closed-loop speed control system, employing a PID (proportional, integral, derivative) control algorithm to adjust the speed of each motor. The PID controller calculates the deviation between the target speed and the feedback speed in real time. It dynamically adjusts the duty cycle of the PWM (pulse width modulation) signal sent to the motor driver by responding to the current deviation with a proportional element, eliminating steady-state error with an integral element, and predicting the trend of deviation changes with a derivative element, thereby correcting the motor's output torque and speed. This control logic ensures that the autonomous mobile chassis 1 can maintain a preset cruising speed in farmland environments with varying loads, and the travel speed is continuously adjustable within the range of 0.5 km / h to 1.5 km / h.

[0036] Furthermore, the energy supply system of the autonomous mobile chassis 1 adopts a hybrid power supply mode. The power module 12 includes a large-capacity battery pack and a fuel generator set electrically connected to it. During operation, the battery pack acts as a direct power source to supply power to the hub motors 5 and on-board electronic equipment to ensure the stability of the output voltage; the fuel generator set acts as a range extender, monitoring the state of charge (SOC) of the battery pack. When the SOC falls below a set lower limit, it automatically starts generating electricity to replenish the battery pack online in real time. This hybrid electric architecture design allows the equipment to meet the requirements of long-term (≥8 hours) continuous operation at night while achieving precise control of low-speed, high-torque operation using electric drive characteristics.

[0037] The autonomous mobile chassis 1 is connected to the driving unit below via an active suspension system. This suspension system integrates an adaptive adjustment mechanism for driving posture, which is used to counteract the impact of uneven farmland furrows on the chassis posture and ensure angular stability in obstacle recognition and ultraviolet radiation. The hardware execution end of this mechanism is the shock absorber 2 that connects the chassis frame and the wheel hub connector 4. The shock absorber 2 uses an electronically controlled variable damping shock absorber (such as a magnetorheological fluid shock absorber or a solenoid valve type hydraulic shock absorber), and its damping characteristics can change continuously in response to externally input electrical signals.

[0038] The sensing end mainly consists of an adaptive adjustment module 3 installed at the geometric center of the autonomous mobile chassis 1. This module contains a built-in six-axis inertial measurement unit (IMU). During operation, the IMU acquires the attitude data of the autonomous mobile chassis 1 relative to the horizontal plane in real time at a high-frequency sampling rate (e.g., 100Hz, 200Hz), specifically including the roll angle of rotation around the longitudinal axis. and pitch angle of rotation about the lateral axis To ensure that the ultraviolet light beam emitted by the UV-C lamp array 9 can penetrate the crop canopy vertically, the system sets the horizontal plane as the reference plane for attitude adjustment, that is, the target roll angle and the target pitch angle are both set to zero.

[0039] After receiving real-time attitude angle data, the main control module 13 first performs a quantitative assessment of the current fuselage tilt and calculates the comprehensive attitude error. This error value reflects the Euclidean distance by which the fuselage attitude deviates from the horizontal reference at the current moment, and its calculation formula is defined as follows: ; in: The roll angle in radians is measured in real time. This is the pitch angle value in radians detected in real time.

[0040] Based on the calculated comprehensive attitude error The main control module 13 uses a PID (proportional-integral-derivative) closed-loop control algorithm to determine the control quantity applied to the damping 2. Given that the PID algorithm is a mature and widely used technology in the field of automatic control, this embodiment uses the following logic for adjustment: the control system first calculates the proportional term, that is, the comprehensive attitude error... Multiplying by a preset proportional gain coefficient generates a restoring damping force proportional to the current tilt degree, quickly suppressing the instantaneous tilt of the fuselage; secondly, the integral term is calculated, which involves integrating and accumulating historical errors during the operation and multiplying by the integral gain coefficient. This step is used to eliminate steady-state tilt errors caused by uneven load distribution (such as center of gravity shift due to fuel consumption); finally, the differential term is calculated, which is the comprehensive attitude error... The rate of change is calculated and multiplied by the differential gain coefficient to predict the trend of fuselage attitude change, and damping is increased in advance to suppress system overshoot and oscillation.

[0041] The main control module 13 linearly superimposes the calculation results of the proportional, integral, and derivative terms to generate the final damping coefficient adjustment command. And convert it into a corresponding current or voltage signal through a digital-to-analog converter to drive the electromagnetic coil or valve body of the shock absorber 2. When one wheel of the autonomous moving chassis 1 drives into a low-lying area, causing the chassis to tilt (i.e. When the load increases, the control system automatically adjusts the damping stiffness of the corresponding and diagonal shock absorbers, changing the support stiffness characteristics of the suspension system, thereby maintaining the horizontal stability of the fuselage during dynamic travel.

[0042] The positioning module 10 and the main control module 13 work together to form the navigation and obstacle avoidance subsystem of the autonomous mobile chassis 1. The positioning module 10 integrates a multi-mode global navigation satellite system receiver, a real-time dynamic differential (RTK) communication unit, and a lidar sensor. To address signal obstruction issues in farmland environments (such as tall windbreaks or inside greenhouses), this embodiment employs a dual-mode navigation logic combining GNSS-RTK absolute positioning and laser SLAM relative positioning.

[0043] In open operating areas, the positioning module 10 preferentially adopts the GNSS-RTK mode. The RTK communication unit receives differential correction data sent by the base station through the antenna 11, and corrects the carrier phase observation values ​​of the satellite positioning data in real time, thereby outputting latitude, longitude, and elevation information with centimeter-level accuracy. The main control module 13 has a built-in coordinate transformation algorithm to convert latitude and longitude data in a geodetic coordinate system (such as WGS-84) into local Cartesian coordinate system data with the farmland operation origin as the reference. The Gauss-Kruger projection or UTM projection algorithms involved in the above coordinate system transformation are well-known technologies in the field of geographic information systems, and their mathematical derivation process will not be detailed here.

[0044] When the positioning module 10 detects a decline in satellite signal quality, i.e., the positioning solution cannot maintain a fixed solution or the horizontal accuracy factor (HDOP) exceeds a preset safety threshold, the system automatically switches to laser SLAM (Simultaneous Localization and Mapping) mode. At this time, the main control module 13 uses environmental point cloud data collected by the lidar to match it with a pre-constructed farmland environment grid map through a point cloud registration algorithm (such as the ICP iterative nearest point algorithm), and calculates the pose change of the autonomous mobile chassis 1 relative to environmental features, thereby achieving continuous positioning under conditions without satellite signals.

[0045] During autonomous cruise, obstacle avoidance logic relies on multi-sensor fusion perception technology. In addition to the identification module 7 for defect detection, the front end of the autonomous mobile chassis 1 is also equipped with a two-dimensional lidar and ultrasonic sensor array for obstacle avoidance. The main control module 13 maps the obstacle information collected by different sensors to a vehicle coordinate system with the chassis center as the origin.

[0046] For obstacle detection, the system employs a multi-level security protection strategy. The main control module 13, based on grid map technology, projects the obstacle point cloud scanned by the LiDAR onto a two-dimensional grid plane and marks the grid status as occupied, free, or unknown. To identify unstructured dynamic obstacles (such as suddenly appearing people or animals), the main control module 13 also integrates visual information from the recognition module 7, utilizing a deep learning object detection algorithm to extract the semantic features and bounding boxes of the obstacles.

[0047] The obstacle avoidance control logic is configured with three distance thresholds: a sensing zone, a deceleration zone, and an emergency stop zone. When the main control module 13 detects an obstacle within the sensing zone of the travel path, it only performs path replanning calculations and attempts to generate a detour trajectory. When the obstacle enters the deceleration zone, the main control module 13 sends a deceleration command to the hub motor 5 and triggers an audible and visual alarm. When the obstacle intrudes into the emergency stop zone (e.g., less than 0.5 meters from the front of the chassis), the system directly cuts off the power supply to the hub motor 5 through the emergency stop circuit at the electrical hardware level, and simultaneously locks the electromagnetic brake to achieve forced braking and ensure operational safety.

[0048] Furthermore, during path tracking control, the main control module 13 employs either pure tracking or Stanley trajectory tracking algorithms. This algorithm calculates the angular velocity commands required for front wheel steering or differential steering in real time based on the lateral and heading errors between the current positioning coordinates and the preset planned path, ensuring high-precision reproduction of the preset operating route stored in the main control module 13. The geometric calculation formulas in the aforementioned trajectory tracking algorithm can be implemented by those skilled in the art by referring to relevant classic robotics textbooks, and will not be elaborated upon here.

[0049] In order to achieve high-fidelity collection of crop disease characteristics in all weather conditions, especially in low-light environments at night, the front arched frame of the autonomous mobile chassis 1 integrates a visual perception subsystem. At the hardware level, this subsystem consists of a recognition module 7, a supplementary lighting module 8, and supporting optical auxiliary components.

[0050] The recognition module 7, acting as the image data acquisition terminal, is rigidly fixed to the top center or side supports of the arched frame of the autonomous mobile chassis 1, with its lens optical axis pointing towards the crop canopy. To suppress motion blur caused by the autonomous mobile chassis 1 during movement, the recognition module 7 uses an industrial-grade global shutter CMOS image sensor instead of a rolling shutter sensor. This sensor is equipped with a dustproof and waterproof housing (IP65 or higher protection rating) designed for farmland operating environments. The lens is a low-distortion wide-angle fixed-focus lens, with a field of view covering the entire canopy width of a single row of crops and a depth of field covering the crop's growth height from root to tip. The recognition module 7 is connected to the main control module 13 via a high-speed data bus (such as a GigE or USB 3.0 interface) to transmit uncompressed raw RGB image data in real time.

[0051] The supplementary lighting module 8 is arranged around the lens of the recognition module 7 in a spatial layout, or symmetrically distributed in a strip on both sides of the recognition module 7 to create a shadowless lighting environment. The core of the light source of the supplementary lighting module 8 is an array of LED beads with a high color rendering index (CRI>90), and its color temperature is set between 5000K and 6000K to simulate natural midday white light. This ensures that the color of the collected disease spots (such as leaf spots, rust spots, and mold) is consistent with the human eye's observation under natural light, avoiding feature extraction errors caused by light source color deviation.

[0052] To eliminate the specular reflection caused by the waxy layer on crop leaves under direct sunlight, an optical diffuser plate is covered on the outer side of the light-emitting surface of the supplementary lighting module 8. This optical diffuser plate is made of milky-white polymethyl methacrylate or polycarbonate material, with a frosted or microlens array treatment on its surface. The point light emitted by the LED is converted into uniform diffused light after passing through the optical diffuser plate, illuminating the surface of the crop leaves, reducing specular noise in the image and preserving the texture details of the leaf surface. In addition, a light intensity sensor is installed near the recognition module 7, facing the external environment, to sense the ambient light intensity. The signal output terminal of the light intensity sensor is electrically connected to the signal input terminal of the main control module 13, transmitting the collected ambient light intensity signal to the main control module 13. The main control module 13 then calculates the signal and sends a PWM dimming command to the drive circuit of the supplementary lighting module 8, providing a hardware feedback basis for subsequent adaptive dimming.

[0053] In the overall layout of the optical system, to prevent ultraviolet light emitted by the rear UV-C lamp array 9 from escaping into the front recognition area and interfering with the white balance parameters of the recognition module 7 or causing image color distortion, a physically flexible light-blocking curtain is installed on the cross-section between the recognition module 7 and the UV-C lamp array 9 in the autonomous moving chassis 1. This flexible light-blocking curtain is made of black light-blocking rubber or high-density opaque fabric. Its top is fixed to the inner wall of the arched frame, and its lower hem hangs naturally to near ground height, forming two independent optical darkroom areas: the front supplementary lighting imaging area and the rear ultraviolet irradiation area. This physical isolation structure ensures the optical environment independence of the imaging system, unaffected by subsequent sterilization lighting.

[0054] The main control module 13 integrates adaptive control logic for the supplementary lighting module 8. This logic aims to address the unstable imaging quality caused by drastic fluctuations in lighting conditions during field operations. The main control module 13 reads the real-time voltage signal from the ambient light intensity sensor via an analog-to-digital converter and converts it into a standard light intensity value. To ensure that the image acquired by the recognition module 7 has a constant brightness baseline, thereby reducing the generalization requirements of the subsequent neural network model for changes in lighting, the system sets an optimal imaging light intensity threshold. This threshold is pre-calibrated based on the photoelectric response curve and signal-to-noise ratio characteristics of the selected CMOS sensor, and represents the minimum luminous flux required to obtain a clear image at the lowest gain setting.

[0055] The control strategy adopts a feedforward control mode, and the main control module 13 controls the ambient light intensity based on real-time monitoring. Calculate the required output brightness for supplementary lighting module 8. To achieve stepless linear adjustment of brightness, the system employs pulse width modulation (PWM) technology to drive the LED array. Definition The duty cycle of the PWM signal output to the LED driver circuit is defined as a normalized range of [0,1]. The main control module 13 calculates the duty cycle at each moment according to the following control law. : ; in: The duty cycle of the PWM drive signal is a dimensionless value. This value directly determines the proportion of the LED driver MOSFET's on-time within one switching cycle, and thus determines the average current flowing through the LED chip. The ambient light intensity value is collected in real time by a light intensity sensor and processed by a low-pass filter, and the unit is lux. The system presets a constant target light intensity value in lux, which must be greater than or equal to the lower limit of saturation light intensity of the identification module 7 within the preset exposure time. This is the light intensity duty cycle mapping coefficient, in lux.-1 This coefficient characterizes the mapping relationship between the light intensity gap and the driving duty cycle. Its value is jointly determined by the luminous efficiency (luminous flux / power) of the LED array and the inverse square attenuation characteristic of the supplementary lighting distance. During the system initialization phase, this coefficient is determined through a calibration procedure to ensure... The corresponding supplemental light intensity can completely cover The requirement for a completely dark environment.

[0056] In the specific circuit implementation, the frequency of the PWM signal generated by the main control module 13 is set to an integer multiple higher than the image acquisition frame rate of the recognition module 7 (e.g., set to above 2000Hz). The purpose of high-frequency modulation is to eliminate light source flicker, prevent the generation of alternating bright and dark stripes in image sensors with rolling shutters or even global shutters, and ensure consistent light energy integration within each exposure cycle. Furthermore, this control logic has a boundary saturation protection function. When the calculated... When the value is greater than 1, the system forcibly limits it to 1. If the image brightness is still insufficient at this time, the main control module 13 will trigger the camera parameter adjustment logic to automatically increase the analog gain of the image sensor or extend the exposure time as a compensation method after the supplementary light brightness reaches the physical limit. When the ambient light intensity is... Greater than or equal to At that time, the system will automatically The power supply to the supplementary lighting module 8 is cut off by setting it to zero, thereby reducing the overall power consumption of the system and extending the lifespan of the LED beads. Through the above strategy, the system can maintain a constant illumination environment in the recognition area in a continuously changing environment from dim twilight to complete darkness at midnight.

[0057] To achieve real-time, high-precision capture of disease features under the limited computing resources of in-vehicle embedded systems, in this embodiment, the edge computing unit within the main control module 13 runs an improved ELS-YOLO target detection model based on structural reparameterization and pruning optimization. This model does not simply call a general black-box algorithm, but is specifically designed for the characteristics of nighttime and low-light imaging conditions in farmland environments, as well as the complex and variable scale of disease textures. Specifically, the model adopts a complete topology including a backbone feature extraction network, a neck feature fusion network, and a decoupled detection head. In the data input and preprocessing stage, the original RGB image acquired by the recognition module 7 is first adjusted to a fixed-size (e.g., 640×640×3) input tensor using a bilinear interpolation algorithm, and the pixel values ​​are normalized to accelerate model convergence. The backbone network uses a lightweight variant of the CSPDarknet structure, which utilizes a cross-stage local network to extract shallow texture features and deep semantic features of crop leaves. The neck network combines the architectural advantages of feature pyramids and path aggregation networks. Through a bidirectional feature pyramid structure that is both top-down and bottom-up, it enhances the ability to locate features of small lesions and solves the technical problem of easy missed detection of early small lesions.

[0058] Considering the physical limitations of mobile hardware in floating-point computing power, this embodiment introduces a Layer Adaptive Amplitude Pruning (LAMP) mechanism. Unlike traditional global threshold pruning, which can easily damage the deep structure of the network, the LAMP mechanism dynamically removes redundant connections with low output contribution from each convolutional layer based on the relative importance score of weight amplitudes. This mechanism achieves a parameter compression rate of over 40% while maintaining the network's global receptive field, and avoids feature collapse caused by over-pruning, ensuring the algorithm's real-time frame rate on embedded devices. In the model training and optimization phases, to address the problem of gradient vanishing when predicted and ground truth boxes do not overlap, leading to convergence failure, and to improve the geometric accuracy of bounding box regression, this embodiment abandons the traditional IoU loss and constructs a CIoU loss function that includes three constraints: overlap, center distance, and aspect ratio. Based on the principle of geometric quantization, the bounding box regression loss... The calculation model is as follows: ; in: The intersection-over-union ratio (IoU) between predicted bounding boxes and ground truth bounding boxes quantifies the degree of spatial overlap between the two regions; its value ranges from [value range missing]. ; Prediction box center point Center point of the actual annotation box The Euclidean distance between them is used to constrain the offset of the positioning center point; The diagonal length of the smallest closure region (smallest bounding rectangle) that can simultaneously cover both the predicted bounding box and the ground truth bounding box is used as a normalization factor. Numerical stability constant (values) ), to prevent the predicted bounding box from completely overlapping with the ground truth bounding box. A division by zero exception occurred. Aspect ratio trade-off coefficient, used to dynamically adjust the weight of the aspect ratio consistency term based on the merits of crossover ratio; Aspect ratio consistency parameter, defined as This is used to measure the geometric similarity between the aspect ratio of the predicted bounding box and the aspect ratio of the ground truth bounding box. and These represent the width and height of the predicted bounding box and the ground truth bounding box, respectively.

[0059] Through backpropagation of the aforementioned loss function, the model can learn the nonlinear mapping relationship from the image pixel space to the disease state space. During the inference phase, the main control module 13 executes a virus severity determination logic based on multi-support evidence fusion. This logic, recognizing that a single dependency confidence level cannot accurately reflect the physical severity of the disease, introduces the proportion of diseased area. As a key quantitative dimension, the percentage of diseased area is... The calculation is the ratio of the total number of disease mask pixels within the detection frame to the total number of pixels in the current crop canopy region of interest (ROI). To prevent false positives caused by illumination noise or background clutter, a confidence threshold is set in the decision logic. Only when the detection confidence level Hierarchical calculations are only initiated when this threshold is exceeded. The system is based on... and The numerical range is used to determine the final disease level using the following piecewise function. ; in: The output disease level identifier has a set of values. , This indicates that there are no diseases or no effective targets have been detected, and the system does not trigger ultraviolet irradiation. This indicates a mild disease; This represents a moderate disease. This represents severe disease and serves as an input index for the subsequent dose control module; The ELS-YOLO model outputs a confidence probability value that the target belongs to the disease category, with a value range of [0,1]. An effective detection threshold (e.g., set to 0.5) is set; detection results below this value are considered background noise or false detections and are filtered out. The percentage of the detected diseased spot area, a dimensionless value; The threshold for distinguishing between mild and moderate disease area (e.g., set to 0.05). The area percentage threshold for distinguishing between moderate and severe diseases (e.g., set to 0.15).

[0060] In this embodiment, the UV-C lamp array 9 does not employ a traditional single high-power light source design in its physical construction. Instead, it adopts an independent addressing matrix structure based on coordinate mapping. Specifically, the UV-C lamp array 9 is distributed in a semi-enclosed manner along the inner wall of the arched frame of the autonomous moving chassis 1, covering the top and both sides of the crop canopy. This array is divided into the following electrical topology: Line × The system comprises independent control units (e.g., a 16×32 matrix), each containing a set (e.g., four LEDs in series) of deep ultraviolet (DUV) LED chips. A multi-channel constant current drive circuit board is also configured to enable independent driving of each control unit. This circuit board, as the core hardware of the execution layer, is connected to the main control module 13 via a high-speed SPI (Serial Peripheral Interface) bus. Each control unit is equipped with an independent metal-oxide-semiconductor field-effect transistor (MOSFET) as an electronic switch, the gate of which is controlled by a dedicated LED driver chip (such as the TLC5940 or a similar chip with PWM grayscale adjustment). This hardware architecture constitutes a concrete implementation of the technical term "independently adjustable light source matrix" as described in the specification, enabling the system to illuminate only the LEDs corresponding to the affected areas, thereby keeping the LEDs in healthy areas off or at low power, thus reducing overall energy consumption and minimizing harm to non-target organisms.

[0061] In terms of optical characteristics, the selected UV-C LED beads have a center wavelength set in the 260nm to 280nm band (with a peak wavelength preferably 275nm). This band is located in the peak region of the DNA / RNA absorption spectrum, which can destroy the genetic material of pathogens and block their replication process. To solve the heat dissipation problem caused by the high energy of ultraviolet photons, the LED bead array adopts a thermoelectric separation packaging process, is soldered onto a copper-based printed circuit board (MCPCB) with a high thermal conductivity, and has aluminum alloy heat sink fins and an active air-cooling fan attached to the back.

[0062] For the generation of drive control signals, the main control module 13 establishes a mapping model from the spatial coordinates of the defect to the LED matrix index. Let the position of the defect target in the vehicle coordinate system be... The parametric equation of the surface on which the LED array is located is: Parameters are introduced here. and These are used as two orthogonal components of the surface coordinate system, replacing the symbols used in other definitions in the aforementioned embodiments. The system calculates the location of the defect using a geometric projection algorithm. On curved surfaces The projection points on the surface are used to determine the set of LED unit indices that need to be activated. For the selected number The output light radiation power of each LED unit is determined by the duty cycle of the drive current. The equivalent average current of the PWM control signal output by the main control module 13 is... The relationship with the digital duty cycle signal is given by the following formula: ; in: Flowing through the Line number The average driving current of the LED unit, measured in amperes (A), directly determines the ultraviolet radiation flux. The peak current set by the constant current drive circuit (e.g., 0.35A) is fixed by the resistance value of the hardware sampling resistor to ensure that the LED operates within a safe current range; The main control module 13 calculates and allocates the pulse width modulation signal duty cycle to the unit according to the disease level, with a value range of [0,1], where 0 indicates off and 1 indicates full power output; Temperature compensation coefficient, with a value range of [0,1]. This coefficient introduces a thermal negative feedback protection mechanism to prevent LEDs from overheating and burning out.

[0063] To determine the temperature compensation coefficient The LED array is evenly distributed with NTC negative temperature coefficient thermistors to monitor the substrate temperature in real time. The protection logic follows a piecewise linear model: ; in: The real-time temperature of the LED substrate, measured by a thermistor, is in degrees Celsius (°C). The set safe operating temperature threshold (e.g., 60°C) represents the upper limit of the temperature range for long-term stable operation of the LED. The set forced shutdown limit temperature (e.g., 85°C) represents the temperature critical point that will cause permanent damage to the chip.

[0064] First case (value 1): When When the temperature is within a safe range, the function output value is 1, indicating that the system does not reduce current and allows the LED to operate at full rated current. and Proceed to full power output.

[0065] The second case (numerical value) ):when When the time is right, the function outputs a linearly decreasing decimal value; this formula calculates the current temperature. Exceeding the safety threshold The amplitude accounts for the entire buffer temperature range The ratio of the driving current is dynamically reduced. This linear cooling strategy aims to maintain operational continuity while preventing further temperature increases, thus achieving thermal balance control.

[0066] The third case (value 0): When When the function output value is 0, it indicates that the temperature has reached a dangerous level. The system forces the compensation coefficient to be set to zero, thereby cutting off the driving current of the corresponding area, causing the LED to immediately turn off and cool down, thus playing a hardware protection role.

[0067] In addition, to ensure personnel safety during operation, a hardware-level safety interlock mechanism is connected in series in the UV-C array control loop. This mechanism includes human infrared sensors (PIR) and chassis tilt sensors installed around the chassis. When the system detects personnel entering a dangerous area within 1 meter of the equipment, or detects that the autonomous mobile chassis 1 has overturned (tilt angle greater than 30 degrees) causing UV light to leak upwards, the interlock circuit will directly cut off the main power input to the LED driver. This safety cut-off action is directly triggered by the hardware logic gate circuit, without going through the software processing of the main control module 13, thus eliminating the potential risk of software crashes. To address the longitudinal installation gap between the identification module 7 (located at the front of the chassis) and the UV-C lamp array 9 (located in the middle and rear of the chassis), To address the resulting spatiotemporal synchronization deviation, the main control module 13 incorporates a set of spatiotemporal synchronization control logic for identification, positioning, and irradiation. This logic does not rely on unstable time delay control but instead establishes a dynamic virtual shift register mechanism based on odometer pulses. This mechanism aims to eliminate synchronization errors caused by fluctuations in the speed of the autonomous moving chassis 1 (such as acceleration, deceleration, slippage, and pauses), ensuring that ultraviolet light can be accurately projected onto the physical coordinates of the damaged areas that have already been traversed.

[0068] This embodiment employs an odometer pulse counting method to achieve accurate longitudinal distance tracking. At the hardware level, the hub motor 5 of the chassis integrates a high-resolution Hall encoder or photoelectric encoder to provide real-time feedback of the accumulated rotation pulse count of the wheels to the main control module 13. The main control module 13 maintains a synchronization system based on pulse counting as a global clock reference. During the image acquisition and coordinate mapping stage, the defect targets captured by the identification module 7 are represented in the image pixel coordinate system as follows: This section uses... Indicates the horizontal pixel coordinates. This indicates the vertical pixel coordinates, distinguishing them from the symbols used in the previous embodiments. The main control module 13 uses a monocular vision inverse perspective transformation algorithm to convert these pixel coordinates into vehicle physical coordinates relative to the center of the lens of the recognition module 7. To simplify calculations and improve real-time performance, assuming the pose is horizontal, the longitudinal relative distance is... The calculation formula is as follows: ; in: The longitudinal horizontal distance of the target point of the defect from the optical center of the identification module 7 in the vehicle coordinate system is expressed in meters (m). The vertical installation height of the optical center of the recognition module 7 above the ground is measured in meters (m). The initial installation tilt angle of the optical axis of the identification module 7 relative to the vertical ground, in radians (rad). The vertical pixel coordinates (row index) of the center point of the lesion in the image plane; The principal coordinates of the optical axis at the longitudinal center of the image sensor (usually half the pixel value of the image height); The equivalent focal length of a camera in the vertical dimension, measured in pixels.

[0069] Obtain the longitudinal distance of the disease relative to the lens. Then, the system calculates the physical travel distance from the defect point to the trigger plane where the UV-C lamp array 9 is located. .set up To identify the fixed longitudinal physical distance between the optical center projection point of module 7 and the starting row of UV-C lamp array 9 (this value has been calibrated and stored in non-volatile memory at the factory), the total travel distance is defined as follows: .

[0070] To convert this physical distance into incompressible encoder pulse increments, the system calculates the target trigger pulse number based on the wheel's kinematic parameters. : ; in: When the recognition module 7 acquires the frame image, the main control module 13 reads the current global pulse count value (integer) from the underlying driver. The physical straight-line distance required to move the diseased area from its current identified location to the ultraviolet-irradiated area, in meters. ; The dynamic diameter of the walking tires of the autonomous mobile chassis 1, in meters (m). The total number of pulses generated by the encoder for one revolution of the wheel depends on the number of encoder lines and the reduction ratio of the gearbox. The tire slip correction factor (dimensionless, typically ranging from 0.95 to 1.0) is used to compensate for positive errors in mileage measurement caused by tire slippage due to soft soil in farmland.

[0071] Within each control cycle (e.g., every 10 milliseconds) of the main control module 13, the system reads the current encoder pulse count in real time. ,and Calculated Including the horizontal coordinates of the disease and the calculated disease level It is encapsulated as a task object. It is then pushed into the FIFO (First In First Out) queue of tasks to be executed, which is managed by the main control module 13.

[0072] Iterate through the first element of the task queue. When the condition is met... This indicates that the UV-C lamp array 9 has physically moved above the previously identified defect location. At this point, the system uses the horizontal coordinates of the task object... Map the corresponding LED bead column index And according to the disease level The corresponding duty cycle instruction is retrieved to drive the specified LED unit to flash. (Horizontal LED column index) The mapping logic is based on the principle of linear interpolation: ; in, The total effective coverage width (in meters) of the UV-C lamp array 9. The lateral width of a single LED control unit (in meters). This indicates a floor operation. Through the above-mentioned synchronization mechanism based on mileage pulses, regardless of whether the autonomous moving chassis 1 is accelerating, decelerating, or pausing, the illumination action is always strictly anchored to the absolute position in physical space, avoiding the displacement of the illumination position caused by changes in vehicle speed.

[0073] To ensure the cumulative radiation dose applied to the crop leaf surface during mobile operations (Unit: The system remains unchanged regardless of fluctuations in the travel speed of the autonomous mobile chassis 1, and can apply differentiated lethal doses for diseases of varying severity. The main control module 13 operates a dynamic dose closed-loop adjustment algorithm. This algorithm, as the underlying logic implementation of the core technical feature of precise variable irradiation, establishes a multivariate control model that couples operating speed, disease severity, and light source characteristics.

[0074] Within each time step of the control cycle, the system executes energy matching logic based on agronomic requirements. The main control module 13 internally stores a dose lookup table based on agronomic experimental calibration, which defines the doses for different disease severity levels. Minimum effective lethal energy density required For example, for mild diseases ( Set a lower antibacterial dose to protect the crop cuticle; for severe diseases ( A high kill dose is set to completely block the spread of pathogens.

[0075] However, the actual energy density projected onto the crop leaf surface is affected by both the instantaneous speed of the vehicle and the output power of the light source. To decouple these two variables, this embodiment constructs the following real-time duty cycle calculation model to solve for the theoretical pulse width modulation signal duty cycle under the current control cycle. : ; in: The calculated theoretical target duty cycle, with a range of values. This value serves as an intermediate variable for subsequent saturation and verification processes. Based on the current disease level (range of values) The target illuminance energy density obtained from the table is in units of ( This reflects the agronomic requirement of applying pesticides as needed; The actual travel speed of the autonomous mobile chassis 1, calculated in real time by the wheel speed sensor, is expressed in units of ( ). This parameter is introduced to counteract the interference of vehicle speed changes on the cumulative dose, that is, to follow the compensation principle of increasing power when the vehicle speed is high and decreasing power when the vehicle speed is low. The UV-C lamp array 9, driven at rated full power (100% duty cycle) at a standard reference distance... The irradiance measured at the location, in units of ( This value is obtained from factory calibration; The effective light spot length of the LED array in the direction of vehicle travel is expressed in ( ). This parameter determines the effective irradiation time of a point on a crop leaf through the irradiated area; Distance attenuation correction factor (dimensionless), which is the vertical distance between the leaf surface and the light bulb due to different crop canopy heights. It will change. According to the inverse square law of optics, this coefficient is calculated as... ,in To calibrate the reference distance, The distance is acquired in real time by an ultrasonic ranging sensor installed on the side of the light panel. When the actual distance is greater than the reference distance, this coefficient is greater than 1, and the system will automatically increase the output power to compensate for transmission loss. Light source aging compensation coefficient (dimensionless) The system records the cumulative working time of the LED beads. And based on the lifetime decay curve fitting formula (e.g. , The coefficient (which is an aging rate constant) is dynamically adjusted to compensate for the decrease in luminous efficiency caused by chip aging.

[0076] The calculated theoretical demand duty cycle Before being sent to the underlying driver, a saturation nonlinearity verification step is required. This step constitutes a coordinated closed-loop control of speed and power to handle situations where the maximum light intensity still cannot meet the dose requirement. The saturation verification and speed reprogramming model is as follows: ; And the final execution duty cycle: ; in: After saturation check logic correction, the main control module 13 sends a new target speed command to the mobile chassis motion controller, in units of ( ; Within the current control cycle, the originally set target command speed or the speed generated by the navigation algorithm, in units of ( The value 1 represents the physical upper limit of the duty cycle, that is, the LED driver is in full power (100%) output state; The first case ( When the theoretically required duty cycle does not exceed the physical limit, the system determines that the current light source power is sufficient. At this time, the new command speed... Maintain the original command speed No speed reduction will be implemented; The second scenario ( When the theoretically required duty cycle exceeds the physical limit, it means that even with the LEDs at full power, a sufficient lethal dose cannot be provided at the current vehicle speed. At this point, based on the principle of energy conservation, the system proportionally reduces the commanded speed. This is done to extend the exposure time to compensate for insufficient power; The actual execution duty cycle is finally sent to the LED driver chip via the hardware interface, with a value range of [0,1]. Minimum function. This operation ensures that regardless of the theoretically calculated value... How large is the actual signal output to the hardware? It is always clamped within the physical limit value of 1 to prevent the drive circuit from being overloaded.

[0077] Through this feedback regulation strategy that combines optical irradiation and mechanical motion, the system achieves millisecond-level power following control at the micro level and speed planning based on operational quality constraints at the macro level, ensuring that the ultraviolet dose applied to the diseased area will not be lower than the agronomically set threshold under any operating condition.

[0078] To ensure full traceability of the operation process and provide a foundation for subsequent agronomic big data analysis, the main control module 13 also includes a data acquisition and storage unit. This unit is not limited to simple log recording but rather constructs a spatiotemporally aligned multimodal data fusion architecture.

[0079] The main control module 13 aggregates heterogeneous data streams from the vision recognition system, motion control system, and actuator feedback system in real time via the CAN (Controller Area Network) bus and internal shared memory mechanism. To address the data dispersion problem caused by inconsistent sampling frequencies of different sensors, the main control module 13 establishes a data alignment layer based on a unified timestamp (Unix Timestamp). This layer encapsulates events occurring at the same time into a standardized multimodal data frame. The mathematical description is as follows: ; in: The data frame was generated using a precise system timestamp, accurate to the millisecond. Geospatial information vectors, which include longitude, latitude, and elevation data obtained from the Global Positioning System; The visual feature information vector contains the hash index value of the disease image and the location information of the detection box in the image coordinate system; The environmental condition information vector includes data on temperature, humidity, and ambient light intensity at the work site; The operation control parameter vector includes the vehicle's instantaneous speed, ultraviolet emission power setting, and current cumulative dosage.

[0080] Regarding the storage strategy for massive amounts of data, considering the fluctuations in network bandwidth in field operations and the physical limitations of vehicle-mounted storage space, this embodiment does not adopt a full data storage strategy. Instead, it introduces a data value assessment algorithm based on multi-dimensional feature weighting. This algorithm is used to calculate the storage priority score of the current data frame in real time. Only when the score exceeds a preset threshold is a high-precision image archiving and cloud synchronization operation triggered; otherwise, only lightweight metadata text is saved. The computational model for the data value assessment algorithm is constructed as follows: ; in: The calculated data frame storage priority score is a dimensionless value. The higher the value, the more typical the disease information contained in the current data frame, or the more drastic the environmental changes, and the higher its archiving value. Currently identified disease level (range of values) This parameter is directly inherited from the output of the aforementioned recognition module 7; The highest level constant in the disease grading standard (set to 3 in this embodiment). Used to normalize the severity of diseases, ensuring that data on severe diseases are retained first; The confidence score output by the object detection neural network model, ranging from [0,1]. Samples with high confidence scores have higher positive sample value for subsequent model iterations and training. Image frame data acquired at the current moment; The keyframe image data that was selected and stored in the previous moment; Image similarity functions (using either SSIM or histogram cosine similarity algorithms) are used to calculate the image similarity of the current frame. Compared to the previous keyframe image The degree of visual similarity, with values ​​ranging from [0,1]; Represents the degree of difference or novelty of image content. This item is used to remove highly repetitive and redundant images (such as consecutive similar images captured when the vehicle is stationary or moving at a slow speed), retaining only frames where the scene changes; Special area indicator function (value is 0 or 1). The system has preset geofence coordinates for key monitoring areas (such as the core area of ​​the experimental field). When the vehicle's location coordinates fall within this area, the function value is 1; otherwise, it is 0. This implements mandatory data retention based on geographic location. The weight coefficients of each evaluation indicator satisfy the normalization condition. In this embodiment, the preferred values ​​are 0.4, 0.2, 0.3, and 0.1.

[0081] The data, after the above filtering, is physically stored using a hierarchical storage mechanism. The underlying layer uses a high-performance key-value database based on LSM-Tree (Log Structure Merging Tree) to store metadata, including structured text information such as time, location, disease type, and job parameters, ensuring millisecond-level write speeds and retrieval efficiency. For unstructured, large-volume data (such as raw disease images), it is stored as a file stream in a specific partition of the vehicle-mounted solid-state drive (SSD), and only its file path index is stored in the database.

[0082] Furthermore, to prevent data corruption due to unexpected power outages, a write-ahead log mechanism is employed during the data writing process. Before flushing the data to the main disk structure, it is first appended to a sequential log file. The underlying implementation code for the aforementioned hierarchical storage and log protection mechanism can be implemented by those skilled in the art using existing embedded database technologies; the specific software stack configuration is well-known in the field and will not be elaborated upon here. This hardware-software integrated data acquisition scheme ensures that the status of each crop being processed is traceable, achieving a digital closed loop from the field to the cloud database.

[0083] To ensure the authenticity and integrity of collected agricultural data and operational records during transmission and storage, and to prevent malicious data tampering or silent data errors due to hardware failure, a chain-based encryption verification mechanism is introduced at the storage layer. This mechanism uses cryptographic algorithms to lock discrete time-series data into an irreversible hash chain.

[0084] Each standard multimodal data frame generated in the aforementioned embodiments is encapsulated as an independent block node. When writing the current data block to the storage medium, not only is the current data payload stored, but a unique digital fingerprint is also calculated and attached. The generation of this digital fingerprint no longer depends solely on the current data, but adopts a recursive chaining approach. Specifically, the system uses a one-way hash function (such as the SHA-256 algorithm) to generate the cryptographic hash value of the current block. The input for calculating this hash value includes a binary stream composed of three parts: the first part is the complete content of the current data frame, covering timestamps, location coordinates, disease image features, and application parameters; the second part is the hash value already generated for the previous data block; and the third part is a random salt value generated by the true random number generator within the main control module 13 or a device-unique hardware key.

[0085] By incorporating the hash value of the previous block into the hash calculation of the current block, a strongly coupled chain dependency is established among all historical data. If an attacker attempts to modify any block in the history, the avalanche effect of the one-way hash function will cause a drastic change in the hash value of that block, leading to the failure of the chain verification logic for all subsequent blocks. Unless the attacker can simultaneously recalculate all hash values ​​from the tampered point to the current point and overwrite the storage, any tampering will be detected by the auditing system in a timely manner. In addition, to prevent rainbow table attacks, a random salt value is introduced to ensure that even if the job data content of different devices is the same, the generated hash chain is unique. To further enhance security, the main control module 13 also includes a Trusted Platform Unit (TPM), which writes the current chain head hash value to a read-only counter or a one-time write-to-store (OTP) area within the TPM at fixed time intervals. During data auditing, the host computer only needs to verify whether the chain relationship is valid and whether the chain head hash matches the record in the TPM to determine whether the data has been tampered with.

[0086] To transform the underlying machine data into decision support information understandable to agronomists, a host computer analysis software is also provided. This software, based on a Geographic Information System (GIS) kernel, offers functions such as operation trajectory playback, generation of disease distribution heatmaps, and evaluation of pesticide application effects. In disease distribution analysis, since the sensors collect discrete disease point coordinates, to visually display the disease outbreak trend across the entire plot, the system employs a kernel density estimation algorithm to transform discrete disease detection events into a continuous disease density field. This algorithm traverses every pixel or grid point within the monitoring area, calculating the superimposed influence of all surrounding disease sample points on that point.

[0087] For any location on the map to be evaluated, the system calculates its Euclidean distance to each collected disease sample point. This distance is then converted into a probability density value using a kernel function (e.g., a standard Gaussian kernel). The kernel function simulates the probability decay characteristic of disease spreading from the center outwards; that is, the closer the distance to the sample point, the greater the density contribution; the farther the distance, the exponentially the contribution decreases. During the overlay calculation, the contribution value of each sample point is multiplied by a weighting coefficient based on the disease level, resulting in higher-level disease points producing a stronger signal response in the final density field. Finally, the weighted contribution values ​​of all sample points to the location to be evaluated are summed and normalized by dividing by the total number of samples and the square of the smoothing bandwidth parameter to obtain the final disease density estimate for that location.

[0088] In the visualization interface, the density estimate is mapped to a color gradient from cool tones (low density) to warm tones (high density), overlaid on the satellite map layer. The smoothing bandwidth parameter in the algorithm determines the richness of detail in the heatmap; a smaller bandwidth results in more obvious local details, while a larger bandwidth produces a smoother overall trend. Through this visualization analysis, farmers can intuitively view high-incidence areas of diseases and key areas for pesticide application after the operation is completed. The generated disease distribution map can be exported as a standard spatial data format and used as input for prescription maps in the next round of operations or for spraying by agricultural drones, achieving closed-loop utilization of operational data.

Claims

1. A method for controlling crop diseases based on precise ultraviolet irradiation, characterized in that, Includes the following steps: The positioning module (10) receives the operation start command and begins to parse the current location information. The main control module (13) controls the hub motor (5) to drive the autonomous mobile chassis (1) to autonomously cruise along the crop ridges. During the journey, the adaptive adjustment module (3) adjusts the damping force of the shock absorption damping (2) according to the real-time collected attitude data. The supplementary lighting module (8) dynamically adjusts the luminous intensity according to the ambient light intensity, the identification module (7) continuously collects crop images, and the main control module (13) runs the detection model to output the pixel coordinates and disease level of the disease target; The main control module (13) uses time synchronization logic to calculate the number of target trigger pulses that the disease target enters the coverage area of ​​the UV-C lamp array (9) based on the real-time travel speed of the autonomous mobile chassis (1) and the physical installation distance between the identification module (7) and the UV-C lamp array (9). When the diseased target reaches the irradiation position, the main control module (13) calculates the required target radiation dose according to the disease level and controls the ultraviolet emitting unit at the corresponding position in the UV-C lamp array (9) to perform fixed-point irradiation. Throughout the entire operation, the operation traceability module (14) synchronously collects operation data, uses encryption technology to generate tamper-proof data blocks, and uploads them through the antenna (11).

2. The method for controlling crop diseases based on precise ultraviolet irradiation according to claim 1, characterized in that, The steps of the supplementary lighting module (8) to dynamically adjust the luminous intensity according to the ambient light intensity include: the main control module (13) acquires the ambient light intensity data, reads the imaging light intensity threshold, and calculates the difference between the imaging light intensity threshold and the current ambient light intensity; The main control module (13) generates a pulse width modulation signal duty cycle based on the difference and drives the LED array of the fill light module (8) to fill light.

3. The method for controlling crop diseases based on precise ultraviolet irradiation according to claim 1, characterized in that, In the step where the main control module (13) calculates the required target radiation dose based on the severity of the disease and controls the ultraviolet emitting units at the corresponding positions in the UV-C lamp array (9) to perform targeted irradiation, a synchronous control logic based on odometer pulses is adopted: The main control module (13) obtains the current global pulse count value fed back by the hub motor (5), and calculates the target trigger pulse number based on the longitudinal horizontal distance of the target in the vehicle coordinate system collected by the identification module (7), combined with the dynamic diameter of the moving wheel (6) and the slip correction coefficient. The main control module (13) pushes the task object containing the target trigger pulse count, horizontal coordinate and disease level into the queue. When the real-time pulse count reaches the target trigger pulse count, the corresponding position of the lamp column index in the UV-C lamp array (9) is mapped according to the horizontal coordinate, and the ultraviolet emission unit at the corresponding position is driven to work.

4. The method for controlling crop diseases based on precise ultraviolet irradiation according to claim 3, characterized in that, The steps for calculating the required target radiation dose include: The main control module (13) finds the minimum effective lethal energy density according to the disease level and calls the duty cycle calculation model. The duty cycle calculation model uses the actual travel speed of the autonomous moving chassis (1) as the speed compensation variable and the vertical distance between the crop surface and the light bead as the attenuation correction variable for calculation. The main control module (13) calculates the theoretical target duty cycle. If the theoretical target duty cycle exceeds the physical limit, the command speed of the autonomous mobile chassis (1) is reduced.

5. The method for controlling crop diseases based on precise ultraviolet irradiation according to claim 1, characterized in that, The steps of the job tracing module (14) synchronously collecting job data and generating tamper-proof data blocks using encryption technology include: Encapsulate the timestamp, location coordinates, disease images, and pesticide application / irradiation data into a current operation data frame; A random salt value is generated, and a one-way hash function is used to generate the cryptographic hash value of the current block. The input for calculating the cryptographic hash value includes the content of the current job data frame, the hash value of the previous data block, and the random salt value.

6. The method for controlling crop diseases based on precise ultraviolet irradiation according to claim 1, characterized in that, The adaptive adjustment module (3) adjusts the damping force of the shock absorption damping (2) according to the real-time collected attitude data as follows: The square root of the sum of the squares of the real-time detected roll angle and pitch angle is used as the comprehensive attitude error. The PID algorithm is used to calculate the control quantity, and the output electrical signal drives the damping (2) to adjust the output damping force.

7. A cross-ridge operation equipment for controlling crop diseases based on precise ultraviolet irradiation, applied to the method for controlling crop diseases based on precise ultraviolet irradiation as described in any one of claims 1-6, characterized in that, The autonomous mobile chassis (1) has an arched cross-ridge structure as its main frame. The autonomous mobile chassis (1) has a suspension system symmetrically arranged on both sides of its bottom. The suspension system has a walking drive unit at its bottom. The autonomous mobile chassis (1) has a positioning module (10) and an antenna (11) at its top. The autonomous mobile chassis (1) has a power module (12) connected to its side. The autonomous mobile chassis (1) is provided with an integrated unit for supplementary lighting and recognition at the front end in the forward direction. The integrated unit for supplementary lighting and recognition integrates a recognition module (7) and a supplementary lighting module (8). The recognition module (7) is located next to the supplementary lighting module (8). The autonomous mobile chassis (1) is provided with a recognition camera (15) on top. The autonomous mobile chassis (1) has UV-C lamp arrays (9) installed on the top and inner walls of the arched frame. The autonomous mobile chassis (1) is equipped with a main control module (13) and an operation traceability module (14).

8. The cross-ridge operation equipment for controlling crop diseases based on ultraviolet precision irradiation according to claim 7, characterized in that, The suspension system includes shock absorber (2) and adaptive adjustment module (3). One end of the shock absorber (2) is set on the autonomous mobile chassis (1), and the other end is set on the adaptive adjustment module (3). The walking drive unit is set at the bottom of the adaptive adjustment module (3). The walking drive unit includes a hub connector (4), with one end of the hub connector (4) located at the bottom of the adaptive adjustment module (3), and the other end of the hub connector (4) being provided with a hub motor (5), with a moving wheel (6) fixedly connected to the output end of the hub motor (5).

9. The cross-ridge operation equipment for controlling crop diseases based on ultraviolet precision irradiation according to claim 7, characterized in that, The UV-C lamp array (9) is composed of multiple independently controlled ultraviolet emitting units arranged in a matrix; a physical flexible light-blocking curtain is provided between the UV-C lamp array (9) and the supplementary lighting module (8), the top of the physical flexible light-blocking curtain is fixed to the inner wall of the arched frame, and the bottom hangs down naturally.

10. The cross-ridge operation equipment for controlling crop diseases based on ultraviolet precision irradiation according to claim 7, characterized in that, The main control module (13) is connected to the hub motor (5), the adaptive adjustment module (3), the identification module (7), the supplementary lighting module (8), the UV-C lamp array (9), the positioning module (10), and the operation traceability module (14) via electrical connection or data bus connection; the power module (12) includes a battery pack and a fuel generator, and the battery pack is connected to each of the above electronic components via electrical connection.

Images