Crop fungal disease control method and device based on ultraviolet rays

Abstract

The invention relates to the technical field of agricultural robots, and discloses a crop fungal disease control method and device based on ultraviolet rays, and the device comprises an energy power system, an autonomous mobile chassis, an ultraviolet irradiation execution system and a multi-source cooperative control system. The multi-source cooperative control system evaluates a disease risk based on the multi-source sensing data and generates an operation instruction; and the ultraviolet irradiation execution system calculates a volume light resistance density index by using canopy three-dimensional scanning data, adaptively adjusts UVC irradiation power, an emission angle and an electric telescopic arm stroke, controls the wind field collaborative ventilation module to generate directional airflow to disturb crop leaves, and realizes coverage of the back surfaces of the leaves and the interiors of canopies in cooperation with ultraviolet rays. The energy power system adopts an oil-electricity hybrid power supply mode, and the energy flow direction is dynamically dispatched to guarantee the operation endurance under the high-load working condition. The problems of shielding and blind areas in physical prevention and control are effectively solved, and the accuracy and the working efficiency of disease prevention and control are improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural robot technology, specifically to a method and device for controlling fungal diseases in crops based on ultraviolet light. Background Technology

[0002] In both greenhouse agriculture and field cultivation, fungal diseases such as powdery mildew and gray mold are major factors affecting crop yield and quality. Long-term reliance on chemical fungicides not only easily leads to pathogen resistance but also causes excessive pesticide residues and environmental pollution. As an alternative to green pest control, physical control technologies that utilize short-wave ultraviolet (UVC) light to disrupt the DNA structure of pathogens are increasingly being applied in agricultural production practices due to their residue-free, broad-spectrum, and highly effective characteristics.

[0003] In existing technologies, ultraviolet (UV) irradiation equipment mostly uses self-propelled trolleys or tracked robots as carriers, carrying fixed light sources to irradiate crops from the side or top. However, because UV radiation follows a linear propagation pattern and has weak penetrating power, when facing crops with high canopy closure, the outer leaves physically block the inner branches and leaves, as well as the undersides of the leaves. This static irradiation method makes it difficult to cover the radiation field to the undersides of leaves and inner canopies, areas prone to disease and hidden, resulting in blind spots in disease control and making it easy for diseases to recur shortly after treatment.

[0004] Meanwhile, conventional pest control equipment typically lacks real-time perception and adaptive parameter adjustment mechanisms for the three-dimensional morphology of crop canopies. During operation, fixed irradiation distances and emission power are often used, failing to adapt to the varying needs of different crop growth stages and planting densities. This can easily lead to problems such as excessive radiation doses causing crop burns due to sparse canopies, or insufficient internal radiation doses due to dense canopies. Furthermore, to ensure sterilization efficiency, ultraviolet light sources often need to maintain high power output, which, combined with the energy consumption of the mobile chassis, places a significant burden on the energy system. Existing pure electric agricultural robots, when performing such high-load operations, are limited by battery energy density, resulting in short battery life and frequent charging requirements that reduce the continuity and efficiency of large-scale farmland operations. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and apparatus for controlling fungal diseases in crops based on ultraviolet light, which solves the problems mentioned in the background section.

[0006] The first aspect of this invention provides a method for controlling crop fungal diseases based on ultraviolet light, applied to a crop fungal disease control robot. The method includes: collecting environmental parameters using a multi-source sensor fusion module, combining this with location information from a positioning and navigation module and historical disease data to generate an input vector; calculating the disease risk level based on the input vector using a disease risk decision module; and triggering a multi-source collaborative control system to generate an operation command when the disease risk level reaches a threshold or when a canopy 3D scanning recognition unit identifies disease characteristics.

[0007] In response to the operation command, the canopy 3D scanning and recognition unit scans the crop canopy to obtain parameters such as canopy height, canopy thickness, and leaf density. Based on these parameters, the multi-source collaborative control system calculates the target emission angle, target irradiation power, and extension stroke of the electric telescopic arm of the UVC intelligent irradiation module, and calculates the fan speed of the wind farm collaborative ventilation module based on the target irradiation power and the preset linkage ratio. The autonomous moving chassis moves forward, and the electric telescopic arm delivers the UVC intelligent irradiation module to the operation position and adjusts the irradiation posture according to the target emission angle. The UVC intelligent irradiation module emits ultraviolet rays at the target irradiation power, and the wind farm collaborative ventilation module generates airflow disturbance to the crop blades based on the fan speed. During and after the operation, the remote operation and maintenance management module monitors the status and uploads data through the dual-mode communication module, updating the disease database and strategies.

[0008] Furthermore, the steps for generating input vectors and calculating disease risk levels include: performing spatiotemporal alignment and normalization processing on environmental parameters, crop status data, and historical disease data to generate normalized input vectors; calculating the fungal disease outbreak rate based on a multivariate logistic regression model, and mapping the fungal disease outbreak rate to a discrete baseline risk level. The canopy 3D scanning and recognition unit integrates a visual sensor and introduces a machine vision-based forced correction mechanism: when the visual sensor detects that the confidence level of a disease phenotype exceeds the lower limit and the area of ​​suspected lesions exceeds the minimum judgment threshold, the visual marker position is deemed valid; risk arbitration is performed based on the principle of prioritizing phenotypic evidence; if the visual marker position is valid, the final execution risk level is locked at no less than the preset minimum level threshold; if the visual marker position is invalid, the larger value between the baseline risk level and the minimum level threshold is taken.

[0009] Further, the steps for calculating the target emission angle include: dividing the canopy point cloud set into continuous differential slices along the direction of travel, and calculating the average canopy height and average canopy thickness of each slice; discretizing the slice space into a three-dimensional voxel grid, counting the number of voxels occupied by the echo, and calculating the volumetric light resistance density index to characterize the leaf density parameter; using a dynamic equilibrium strategy based on the volumetric light resistance density index to calculate the optimal light coupling distance of the UVC intelligent irradiation module: shortening the distance when the volumetric light resistance density index approaches 1, and increasing the distance when the volumetric light resistance density index approaches 0; and calculating the target emission angle based on the principle of radiation centroid alignment, according to the height of the vertical geometric center of the canopy slice and the height of the rotation axis of the UVC intelligent irradiation module.

[0010] Furthermore, the calculation of the target irradiation power adopts an exponential penetration compensation model: First, based on the real-time disease risk level, a graded gain model is used to calculate the target radiation dose; combined with the travel speed of the autonomous mobile chassis, the reference emission power is inversely calculated; an exponential term with the volumetric light resistivity density index as the independent variable is introduced to perform nonlinear compensation on the reference emission power to generate the target irradiation power. If the target irradiation power exceeds the hardware rated power, the power is truncated to the rated value, and a speed degradation request signal is generated and fed back to the autonomous mobile chassis to maintain a constant cumulative radiation dose by reducing the travel speed. The calculation of the fan speed of the wind field coordinated ventilation module includes: establishing an aerodynamic impedance model of the stratum to convert the volumetric light resistivity density index of the canopy into an ideal penetration wind speed; following the energy matching principle, calculating the target fan speed, which is positively correlated with the target irradiation power; superimposing a sinusoidal perturbation on the target fan speed, the frequency of which is dynamically determined by the travel speed of the autonomous mobile chassis.

[0011] Furthermore, in the autonomous mobile chassis movement process: the dual-mode drive unit constructs a terrain travel resistance index based on soil volumetric moisture content and local slope angle; the terrain travel resistance index is compared with a mode switching threshold, and when the index is less than or equal to the threshold, it switches to wheel drive mode, and when it is greater than the threshold, it switches to track drive mode; the vehicle roll angle and pitch angle are monitored in real time by the driving posture adaptive adjustment unit, and the target vertical adjustment amount of the suspension support point is calculated; a semi-active ceiling damping control strategy is adopted to adjust the suspension damping coefficient in real time according to the vehicle's vertical absolute speed and the wheel's vertical speed to keep the UVC intelligent irradiation module's irradiation reference plane in a horizontal state.

[0012] Furthermore, the steps for updating the disease database and strategies include: constructing an online health assessment model based on electrothermal-optical multi-physics coupling; calculating the intrinsic photoelectric conversion efficiency of the light-emitting units in the UVC intelligent irradiation module normalized to the standard temperature; and determining the aging status of the devices by combining the equivalent dynamic resistance drift rate; calculating the lifetime loss factor based on the Arrhenius model and combining real-time junction temperature and sampling current to predict the remaining lifetime of core components; dynamically adjusting the power compensation strategy for subsequent operations based on the remaining lifetime and device aging status; and comparing the predicted disease level before the operation with the actual disease hotspots identified during the operation. If the actual disease level is found to be significantly higher than the predicted level, the initial risk weight of the next round of operations for the plot is adjusted, and the disease risk decision model is corrected.

[0013] A second aspect of this invention provides an ultraviolet-based device for controlling crop fungal diseases, including a crop fungal disease control robot. The robot comprises an energy and power system, an autonomous mobile chassis, an ultraviolet irradiation execution system, and a multi-source collaborative control system. The energy and power system provides electrical power; the autonomous mobile chassis carries the energy and power system, the ultraviolet irradiation execution system, and the multi-source collaborative control system; the ultraviolet irradiation execution system is positioned above the autonomous mobile chassis and is used for physical control of the crop canopy; the multi-source collaborative control system is communicatively connected to the energy and power system, the autonomous mobile chassis, and the ultraviolet irradiation execution system.

[0014] Furthermore, the energy and power system includes a fuel generator, an energy storage module, a solar charging module, and a load monitoring module; the fuel generator and energy storage module are electrically connected to form a hybrid power supply circuit; the load monitoring module is used to collect real-time power demand data of the entire machine and state-of-charge data of the energy storage module. The autonomous mobile chassis includes a dual-mode drive unit, a driving posture adaptive adjustment unit, a positioning and navigation module, and a road condition recognition module; the dual-mode drive unit includes an all-wheel drive mechanism and a track drive mechanism, used to switch drive modes according to terrain parameters; the driving posture adaptive adjustment unit is connected to the suspension system. The ultraviolet irradiation execution system includes a canopy 3D scanning and recognition unit, a UVC intelligent irradiation module, an electric telescopic boom, and a wind farm coordinated ventilation module. The multi-source collaborative control system includes a multi-source sensor fusion module, a disease risk decision-making module, a dual-mode communication module, and a remote operation and maintenance management module; the dual-mode communication module has a built-in dynamic routing algorithm based on link quality and service demand matching, calculates the transmission cost function of each link based on broadband link quality scores, narrowband link quality scores, and feature vectors of service data flows, and switches between broadband and narrowband networks according to the minimum transmission cost function.

[0015] This invention provides a method and apparatus for controlling fungal diseases in crops based on ultraviolet light. It has the following beneficial effects: 1. This invention establishes a linkage mechanism between a wind-coordinated ventilation module and a UVC intelligent irradiation module. By utilizing the disturbance field generated by directional airflow, it forcibly induces crop leaves to turn and tremble. The light-wind coordinated operation mode can expose the hidden parts of the underside of the leaves and deep within the canopy to the ultraviolet radiation field. This effectively overcomes the shortcomings of traditional static ultraviolet irradiation technology, which has poor penetration and can only cover the surface of the crop. It improves the inactivation rate of fungal diseases latent on the underside of leaves and in the inner canopy, solves the problems of shading and blind spots in ultraviolet physical control, and enhances the thoroughness of disease control.

[0016] 2. This invention utilizes a canopy 3D scanning and recognition unit to acquire the geometric morphology parameters of crops in real time and calculates the volumetric light resistance density index. The multi-source collaborative control system then uses this information to calculate the optimal irradiation distance, emission power, and fan speed. Based on closed-loop control logic with environmental perception, the robot can automatically match the radiation dose to crops at different growth stages and planting densities. This avoids leaf burn caused by excessive power or too close distance, and also prevents disease residue caused by insufficient dose. It achieves adaptive and precise adjustment of operating parameters, balancing the prevention and control effect with crop safety.

[0017] 3. This invention integrates a hybrid energy power system that combines fuel-fired power generation, photovoltaic energy replenishment, and energy storage regulation. Through a load monitoring module, the power flow of the entire machine is scheduled in real time. Under high-load conditions driven by UVC irradiation and strong winds, the system can dynamically intervene in fuel-fired power generation based on the charge state of the energy storage unit. This effectively solves the problems of short endurance and large voltage fluctuations of pure electric agricultural robots when performing high-power physical pest control tasks. It meets the needs of long-term continuous operation in large-scale farmland, breaks through the endurance bottleneck of high-energy-consuming operations, and ensures the continuity and stability of field operations. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system architecture of the present invention; Figure 3 This is a three-dimensional view of the tire drive of the crop fungal disease control robot of the present invention; Figure 4 This is a three-dimensional view of the tracked drive robot for controlling fungal diseases in crops according to the present invention.

[0019] Among them, 100, Crop Fungal Disease Control Robot; 110, Energy and Power System; 111, Fuel Generator; 112, Energy Storage Module; 113, Solar Charging Module; 114, Load Monitoring Module; 120, Autonomous Mobile Chassis; 121, Dual-Mode Drive Unit; 122, Adaptive Driving Posture Adjustment Unit; 123, Positioning and Navigation Module; 124, Road Condition Recognition Module; 130, Ultraviolet Irradiation Execution System; 131, Canopy 3D Scanning and Recognition Unit; 132, UVC Intelligent Irradiation Module; 133, Electric Telescopic Arm; 134, Wind Farm Collaborative Ventilation Module; 140, Multi-Source Collaborative Control System; 141, Multi-Source Sensor Fusion Module; 142, Disease Risk Decision Module; 143, Dual-Mode Communication Module; 144, Remote Operation and Maintenance Management Module. 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 Figure 1 -Appendix Figure 4 This invention provides an ultraviolet-based crop fungal disease control device, including a crop fungal disease control robot 100, which includes an energy power system 110, an autonomous mobile chassis 120, an ultraviolet irradiation execution system 130, and a multi-source collaborative control system 140.

[0022] The power system 110 provides electrical support for the movement and operation of the crop fungal disease control robot 100. The power system 110 includes a fuel generator 111, an energy storage module 112, a solar charging module 113, and a load monitoring module 114. The fuel generator 111 and the energy storage module 112 are electrically connected to form a hybrid power supply circuit. The solar charging module 113 is installed on the surface of the robot and is configured to convert solar energy into electrical energy and store it in the energy storage module 112. The load monitoring module 114 is configured to collect real-time power demand data of the entire robot and state-of-charge data of the energy storage module 112.

[0023] The autonomous mobile chassis 120 carries the energy and power system 110, the ultraviolet irradiation execution system 130, and the multi-source collaborative control system 140. The autonomous mobile chassis 120 includes a dual-mode drive unit 121, a driving posture adaptive adjustment unit 122, a positioning and navigation module 123, and a road condition recognition module 124. The dual-mode drive unit 121 is equipped with both an all-wheel drive mechanism and a tracked drive mechanism, capable of switching drive modes according to terrain parameters. The driving posture adaptive adjustment unit 122 is connected to the suspension system, maintaining the robot's level by adjusting damping. The positioning and navigation module 123 integrates a global navigation satellite system receiver and a real-time positioning and mapping radar for acquiring spatial coordinates. The crop fungal disease control robot 100 can also be configured with a single chassis, allowing it to be used as two independent robots depending on whether it uses all-wheel drive or tracked drives.

[0024] An ultraviolet (UV) irradiation system 130 is mounted above an autonomous mobile chassis 120 and configured for physical control of crop canopy. The UV irradiation system 130 includes a canopy 3D scanning and recognition unit 131, a UVC intelligent irradiation module 132, an electric telescopic arm 133, and a wind-coordinated ventilation module 134. UVC refers to short-wave ultraviolet (Ultraviolet C). The canopy 3D scanning and recognition unit 131 is configured to collect data on the height, thickness, and leaf density of the crop canopy. The electric telescopic arm 133 is connected to the UVC intelligent irradiation module 132 and is used to adjust the irradiation distance. The wind-coordinated ventilation module 134 is located on the side of the UVC intelligent irradiation module 132 and is used to generate directional airflow.

[0025] The multi-source collaborative control system 140 is communicatively connected to the energy and power system 110, the autonomous mobile chassis 120, and the ultraviolet irradiation execution system 130. The multi-source collaborative control system 140 includes a multi-source sensor fusion module 141, a disease risk decision-making module 142, a dual-mode communication module 143, and a remote operation and maintenance management module 144. The multi-source sensor fusion module 141 is used to collect environmental data, crop status data, and historical disease data.

[0026] When the crop fungal disease control robot 100 performs control operations, the multi-source sensor fusion module 141 collects environmental parameters of the work area in real time, combines them with the location information obtained by the positioning and navigation module 123 and pre-stored historical disease data to generate an input vector. The disease risk decision module 142 receives this input vector and calculates the disease risk level according to a preset algorithm model. When the disease risk level reaches the preset operation trigger threshold, or when the visual sensor in the canopy 3D scanning and recognition unit 131 identifies diseased characteristics in the crop phenotype, the multi-source collaborative control system 140 generates an operation command.

[0027] Upon receiving the operation instruction, the canopy 3D scanning and recognition unit 131 performs a feature parameterized scan of the crop canopy to obtain canopy height, canopy thickness, and leaf density parameters. Based on these canopy feature parameters, the multi-source collaborative control system 140 calculates the target emission angle, target irradiation power, and extension stroke of the electric telescopic arm 133 for the UVC intelligent irradiation module 132. Simultaneously, based on the target irradiation power, the multi-source collaborative control system 140 calculates the fan speed of the wind farm collaborative ventilation module 134 according to a preset linkage ratio.

[0028] The autonomous mobile chassis 120 travels along the planned path. The dual-mode drive unit 121 selects either wheel drive mode or track drive mode based on the ground slope and soil moisture data fed back by the road condition recognition module 124. The driving posture adaptive adjustment unit 122 monitors the body tilt angle in real time and dynamically adjusts the suspension damping when tilt is detected to keep the irradiation reference plane of the UVC intelligent irradiation module 132 in a horizontal state.

[0029] During operation, the electric telescopic boom 133 delivers the UVC intelligent irradiation module 132 to the designated operating position. The UVC intelligent irradiation module 132 emits ultraviolet rays at the calculated target irradiation power. The wind farm coordinated ventilation module 134 is activated simultaneously, and the resulting airflow disturbs the plant blades, allowing ultraviolet rays to cover the back of the blades and the interior of the canopy.

[0030] During operation, the load monitoring module 114 monitors the sum of the drive power and the irradiation power in real time. When the total load power and the remaining power of the energy storage module 112 do not meet the preset balance conditions, the energy power system 110 controls the fuel generator 111 to start and adjust the output power to supplement the power supply to the energy storage module 112 and maintain the continuity of operation.

[0031] The remote operation and maintenance management module 144 monitors the operating status of each module in real time through the dual-mode communication module 143. When the luminous efficiency of the UVC intelligent irradiation module 132 is detected to be lower than the preset efficiency threshold, the remote operation and maintenance management module 144 generates a maintenance warning signal and sends it to the remote terminal. After the operation is completed, the crop fungal disease control robot 100 returns to the docking point and uploads the operation data.

[0032] In one preferred embodiment, considering the complex farmland environment and frequent relocation and transportation, in order to solve the problems of protecting precision optical components and occupying equipment space, the present invention introduces an irradiation unit storage function in the ultraviolet irradiation execution system (130). Specifically, the body structure of the crop fungal disease control robot (100) has a semi-cylindrical cavity, and a high-precision electric guide rail is configured inside. When the multi-source collaborative control system (140) determines that the equipment is in a non-operational state (such as standby, relocation, long-distance transportation, or low battery return), it will automatically trigger a storage command. At this time, the electric telescopic arm (133) first resets, and then the lamp array of the UVC intelligent irradiation module (132) smoothly retracts to the inside of the body under the traction of the electric guide rail. The shape of the lamp array matches the curvature of the semi-cylindrical cavity, and after retraction, it can be precisely folded and attached to the inner wall of the semi-cylindrical cavity of the body. After being fully stored, there are no exposed lamp structures on the outside of the robot body.

[0033] The aforementioned storage design serves two purposes. First, the outer shell of the robot body forms a robust physical shield, effectively preventing the expensive UVC lamp beads from breaking or being damaged by bumps or scratches from tree branches during the robot's travel in complex terrain or during loading and unloading. Second, the retracted irradiation module significantly reduces the lateral space occupied by the entire machine (i.e., reduces the width of the device's outline), greatly improving the robot's ability to pass through narrow field roads and reducing the space requirements when stored in warehouses or transport vehicles.

[0034] This invention also provides a method for controlling fungal diseases in crops based on ultraviolet light, comprising the following steps: Step S100: In this embodiment, during the operation initiation phase, the multi-source collaborative control system 140 utilizes the multi-source sensor fusion module 141 to construct a four-dimensional input vector containing real-time environmental parameters, crop status parameters, historical disease parameters, and geographical location parameters. The disease risk decision module 142 calculates the probability of disease occurrence based on a multi-source data weighted fusion algorithm and maps it to a discrete risk level. When the calculated risk level reaches the warning threshold, or when the visual sensor in the canopy three-dimensional scanning recognition unit 131 identifies disease-related characteristics in the crop phenotype, the multi-source collaborative control system 140 determines that there is a need for prevention and control in the current area, generates an operation trigger command, and locks onto the target area.

[0035] Step S101: The multi-source sensor fusion module 141 performs the functions of data aggregation and standardization, mapping the sensor signals from the physical world to the standard input of the digital algorithm space. Addressing the asynchronicity and dimensional differences of heterogeneous sensor data in the spatiotemporal dimensions, the multi-source sensor fusion module 141 executes the following processing flow: Acquisition of environmental and historical parameters: The multi-source sensor fusion module 141 acquires the ambient temperature. relative humidity and instantaneous wind speed Three key physical quantities are used, and when numerical anomalies are detected, the mean of a sliding window is used to fill in the gaps, forming an environmental feature vector. Simultaneously, the multi-source sensor fusion module 141 reads the current coordinates output by the positioning and navigation module 123. Using this as the core of the search, the system searches the existing local historical database for historical disease frequencies within a preset radius. Generate historical disease feature vectors Quantification of crop canopy phenotypes: A three-dimensional canopy scanning and identification unit 131 uses lidar to calculate canopy closure. The canopy closure is calculated based on the transmittance of laser point clouds within a vertically projected cylinder; simultaneously, the canopy 3D scanning and recognition unit 131 calculates the green index using RGB images acquired by a visual sensor. The two parameters mentioned above constitute the crop state feature vector. The multi-source sensor fusion module 141 employs a nearest neighbor timestamp matching method to align sub-vectors based on environmental data timestamps. To eliminate the interference of different physical dimensions on the weights of subsequent decision-making algorithms, the multi-source sensor fusion module 141 constructs a normalized four-dimensional input vector. Its calculation model is as follows: ; in, , and These represent the measured raw data vectors of environment, crop status, and historical diseases after spatiotemporal alignment; , , , , and These are the statistical extreme value vectors or theoretical boundary values ​​of the corresponding parameters; To prevent extremely small positive numbers with a denominator of zero; It is a spatial coordinate vector, which in this embodiment is only used as a spatial index identifier and does not participate in the numerical normalization operation.

[0036] Step S102: Disease risk decision module 142 receives normalized input vector Geographic location parameters are separated through feature decoupling. Extract the remaining scalar elements with definite physical strength and recombine them into a six-dimensional computational feature vector. Six-dimensional eigenvector calculation Subsequently, the disease risk decision-making module 142 performs nonlinear calculations based on a multivariate logistic regression model to simulate the outbreak characteristics of fungal diseases under critical conditions. The specific probability calculation formula is as follows: ; in, This represents the output fungal disease outbreak rate, with a value range of [value range missing]. ; Calculate the eigenvectors for six dimensions; This is the sensitivity weight vector, with a dimension of 6×1, representing the contribution weight of each physical factor to the outbreak of the disease. These are the bias parameters of the model; It is a natural constant. Regarding the sensitivity weight vector... With bias parameters The determination of the model is achieved through an offline supervised learning process in this embodiment. That is, historical sensor data is used as samples, expert diagnosis results are used as labels, binary cross-entropy is used as the loss function, and the model is iteratively optimized through the stochastic gradient descent algorithm until the model converges. This is well known to those skilled in the art, and the specific training formula will not be described in detail here.

[0037] Step S103: To address the potential lag in single-environment prediction models, the disease risk decision-making module 142 introduces a deterministic forced correction mechanism based on machine vision, constructing a dual verification system of probabilistic prediction and phenotypic empirical analysis. The probabilistic discretization disease risk decision-making module 142, based on the economic threshold principle, uses a piecewise quantization function to convert continuous disease risk probabilities... Baseline risk level mapped to integer form : ; in, The preset probability threshold is used. Simultaneously with phenotypic verification and final arbitration, the canopy 3D scanning recognition unit 131 runs a lightweight convolutional neural network model to extract features and classify crop images. When the canopy 3D scanning recognition unit 131 detects that the confidence level of the disease phenotype exceeds the lower confidence limit and the area of ​​suspected lesions exceeds the minimum judgment threshold, it sets a Boolean visual flag. Set to 1 if phenotypic evidence is prioritized and set to 0 otherwise. Finally, the disease risk decision-making module 142 performs risk arbitration based on the principles of prioritizing phenotypic evidence and maximizing risk coverage, calculating the final execution risk level. : ; in, This is the minimum threshold for triggering the actuator (set to 3 in this embodiment). This formula ensures that when the vision sensor directly captures the defect feature (…), the minimum threshold is reached. When this happens, regardless of the probability predicted by the environmental model, the final execution risk level will be determined. It is locked at least at level 3, thereby forcibly triggering the multi-source collaborative control system 140 to generate operation instructions; if the risk level calculated by the environmental model is higher, the higher level is maintained to match the corresponding prevention and control parameters.

[0038] Step S200: In this embodiment, in response to the operation trigger command, the canopy 3D scanning and recognition unit 131 activates the lidar and vision components to scan the crop, reconstructing 3D geometric features such as canopy height, thickness, and leaf density. The multi-source collaborative control system 140 establishes a mapping relationship based on these feature parameters, and reversely calculates the optimal emission angle of the UVC intelligent irradiation module 132 (adapting to canopy morphology), target irradiation power (adapting to disease level and leaf density), and the extension stroke of the electric telescopic arm 133 (adapting to the optimal light distance), thus completing the adaptive parameter preset of the actuator.

[0039] Step S201: The canopy 3D scanning and recognition unit 131 does not only output the raw sensor data stream, but also transforms the unstructured lidar echo signal into structured agronomic morphology parameters through its built-in edge computing node, thereby providing precise spatial constraints for the path planning and power control of the UVC intelligent irradiation module 132. The spatiotemporal registration of the point cloud data responds to the operation instructions issued by the multi-source collaborative control system 140, and the canopy 3D scanning and recognition unit 131 synchronously triggers the data acquisition interface. Given that the lidar data is based on the sensor's local coordinate system, while the autonomous mobile chassis 120's movement is based on the geodetic coordinate system, to eliminate measurement errors caused by installation position offsets, the canopy 3D scanning and recognition unit 131 uses a rigid body transformation algorithm to convert the original point cloud coordinates... Mapped to a unified vehicle reference coordinate system: ; in, The coordinates of the point cloud in the vehicle's reference coordinate system after conversion; It is a 3×3 rotation matrix used to correct the installation attitude angle of the lidar relative to the autonomous mobile chassis 120; The translation vector is 3×1, representing the displacement of the lidar optical center relative to the geometric center of the autonomous mobile chassis at 120°. After coordinate correction, the system applies a pass-through filtering algorithm to extract the region of interest, filtering out ground weeds and distant background, and generating a point cloud set of target crops. To accommodate the differences in crop growth along the vertical axis, the canopy 3D scanning and recognition unit 131 uses a differential slicing method to calculate canopy geometric parameters. The system then uses a point cloud dataset... Divide along the direction of travel (X-axis) into sections with a width of (e.g., 10cm) continuous differential slices. For the first... Calculate the average canopy height from 1 slice. With average canopy thickness : ; ; in, Indicates the first The 95th percentile value of the effective point cloud in the vertical direction in each slice is used, and the quantile value is used instead of the absolute maximum value to suppress outlier noise interference. The pre-defined ground reference height; and These represent the coordinates of the farthest and nearest boundaries of the point cloud slice in the horizontal direction, respectively.

[0040] The leaf density, calculated based on the voxelized leaf density light resistance index, directly determines the ultraviolet light penetration efficiency. In this embodiment, the canopy 3D scanning and recognition unit 131 discretizes the slice space into a 3D voxel grid (side length...). ), count the number of voxels occupied by lidar echoes Volumetric light resistivity index The calculation formula is as follows: ; in, The dimensionless value of the normalized value (takes...) A higher value indicates a denser canopy. The volume occupied by entities representing crop branches and leaves; The total volume envelope of the slice ( ); To calculate the effective minimum volume threshold, in order to prevent division by zero errors; This is the canopy saturation coefficient (e.g., 0.7), used to correct for the physical fact that natural crops cannot achieve 100% canopy filling. This index... The optical penetration resistance characteristics required for physical prevention and control have been fully quantified. Step S202: UVC emission geometric parameter mapping. The multi-source collaborative control system 140 dynamically calculates the position parameters of the actuator based on the canopy three-dimensional features reconstructed in step S201, to ensure that the light radiation field of the UVC intelligent irradiation module 132 covers the crop target with the best geometric coupling efficiency. Optimization of operating distance based on illumination and safety constraints. In order to balance radiation intensity and coverage range while ensuring physical safety, the multi-source collaborative control system 140 adopts a method based on the leaf density light resistance index. The dynamic balancing strategy calculates the optimal optical coupling distance. : ; in, The preset minimum mechanical safety anti-collision clearance (e.g., 100mm); This is the penetration compensation coefficient. When the canopy is dense ( When the canopy is sparse, the system prioritizes shortening the distance to enhance penetration energy; when the canopy is sparse ( When the system is in operation, the distance is appropriately increased to expand the light spot coverage area and improve work efficiency. The inverse kinematics solution of the electric telescopic boom, combined with the spatial position information of the canopy surface, by the multi-source cooperative control system 140, calculates the target telescopic stroke of the electric telescopic boom 133. This process introduces a saturation constraint function. To prevent mechanical overload: ; in, Fixed lateral offset for mounting base of electric telescopic boom 133; and These represent the minimum and maximum extension lengths for mechanical limiting, respectively. Radiation centroid alignment and emission angle calculation: To ensure uniform UVC beam coverage of the canopy height, the multi-source cooperative control system 140 calculates the target emission angle based on the radiation centroid alignment principle. : ; in, The vertical geometric center height of the canopy slice; The height of the rotation axis of the UVC intelligent irradiation module 132; This represents the actual horizontal remaining distance of the photon upon reaching the canopy surface. To prevent division by zero errors for extremely small positive numbers.

[0041] Step S203: The multi-source collaborative control system 140 establishes a mapping relationship from biological lethal dose to equipment electrical power and calculates the driving power of the UVC intelligent irradiation module 132 in real time to ensure effective inactivation of pathogens and pathogen spores.

[0042] The multi-source collaborative control system 140 calculates the target radiation dose based on the real-time disease risk level using a graded gain model. (Unit: J / m) 2 ): ; in, A baseline lethal dose threshold set for a specific pathogen; The current risk level; The highest risk level defined for the system (must ensure) ); This represents the risk gain coefficient. The model implements an on-demand irradiation logic that dynamically fluctuates with disease pressure.

[0043] The multi-source collaborative control system 140 combined with the autonomous mobile chassis 120's travel speed Inversely calculate the reference transmit power An exponential penetration compensation model is introduced to overcome the extinction effect inside the canopy and generate the target irradiation power. : ; ; in, The effective coverage width of the beam; The overall efficiency coefficient for photoelectric conversion; This refers to the pulse duty cycle. and This is a numerical protection constant. In the power correction formula, The penetration resistance coefficient (similar to the extinction coefficient), the exponential term It is used for nonlinear compensation of light energy attenuation caused by shading by branches and leaves.

[0044] If calculated Exceeding the hardware's rated power The multi-source collaborative control system 140 not only cuts off the power to It also generates a speed degradation request signal and feeds it back to the motion control unit. The autonomous mobile chassis 120 reduces its travel speed according to the principle of energy conservation and compensates for the insufficient power limit by extending the exposure time, thus forming a dual closed-loop control strategy of power and speed to ensure that the cumulative radiation dose meets the prevention and control standards under any operating conditions.

[0045] Step S300: In this embodiment, the autonomous mobile chassis 120 automatically switches between all-wheel or tracked drive modes based on soil moisture and slope data, and dynamically adjusts the suspension damping through the driving posture adaptive adjustment unit 122 to maintain the machine's level. As the UVC intelligent irradiation module 132 starts operating at a preset power, the wind farm coordinated ventilation module 134 starts simultaneously, with its fan speed and UVC irradiation power coupled and controlled in real time through a preset linkage coefficient. The resulting directional airflow disturbs the plant's blades, and combined with multi-angle ultraviolet irradiation, achieves full coverage protection of the back of the blades and blind spots inside the canopy.

[0046] Step S301: The multi-source cooperative control system 140 analyzes multi-dimensional environmental perception data to perform real-time closed-loop control of the power output mode of the dual-mode drive unit 121 and the suspension parameters of the driving posture adaptive adjustment unit 122, in order to solve the problems of passage and image stability in unstructured farmland environments. Intelligent switching of drive modes based on ground mechanics: To maintain optimal traction efficiency on soil surfaces with different bearing capacities, the multi-source cooperative control system 140 adjusts the drive mode based on soil volumetric water content. With local slope angle Constructing the terrain traffic impedance index : ; in, The threshold for soil saturation moisture content; The soil shear strength attenuation index (usually taken as 1.5 to 2.0). and These are the weighting coefficients for humidity and slope, respectively. The multi-source collaborative control system 140 will... With mode switching threshold Comparison: When When, switch to wheel drive mode; when When the vehicle is in tracked drive mode, the system switches to tracked drive mode. Hysteresis comparison logic is incorporated into the system; switching is only executed after the value has continuously exceeded a threshold for a certain time window to prevent frequent maneuvers. The six-DOF reverse leveling, positioning, and navigation module 123 outputs the vehicle's roll angle. With pitch angle The multi-source cooperative control system 140 aims to maintain the UVC mounting surface level, and calculates the... Target vertical adjustment of each suspension support point : ; in, The geometric coordinates of the suspension point relative to the center of mass; The stroke saturation constraint function limits the adjustment amount to the mechanically permissible range. Within the range. The system sets the attitude dead zone threshold. Active leveling is triggered only when the angle deviation exceeds the threshold. The vibration frequency response modulation multi-source collaborative control system 140 for suspension damping adopts a semi-active ceiling damping control strategy, based on the vehicle's vertical absolute velocity. perpendicular velocity of the wheel Real-time modulation damping coefficient : ; Applying high damping when relative suspension motion exacerbates vehicle body vibration. Apply low damping during road impact transmission This ensures the smooth operation of the UVC module on complex ridges.

[0047] Step S302: The multi-source collaborative control system 140, which integrates wind field and solar power control, utilizes the spatiotemporal coupling effect of aerodynamic field and photon radiation field to induce blade rotation through directional airflow, thus solving the static shading problem. Based on the reference wind speed setting of canopy impedance, the multi-source collaborative control system 140 establishes a layered aerodynamic impedance model and sets the volumetric light resistance density index. Transform into ideal penetrating wind speed : ; in, The minimum critical wind speed that induces blade flutter; This is the wind resistance compensation coefficient; This is the parameter normalization constant; The flow nonlinearity index is used. The solar-wind energy coupling calculation follows the energy matching principle, i.e., high-power solar illumination combined with high-intensity airflow. The multi-source collaborative control system 140 calculates the target fan speed of the wind field collaborative ventilation module 134. : ; in, Target power for UVC; This refers to the gain coefficient for light-wind linkage. The impeller diameter; For transportation efficiency; The rotational speed saturation constraint function is used. To induce blade resonant overturning, the system superimposes a sinusoidal disturbance onto the target rotational speed, generating a real-time dynamic command. : ; in, The target pulse frequency is determined by the travel speed. Dynamically determined: Ensure that there is light in the window. The blades rotate multiple times during the sweep. : Window flipping coverage factor, a preset constant, representing the theoretical number of flips the blade needs to complete during the robot's journey through a window length; Robot movement speed. Wind turbine load and energy consumption boundary management system calculates and predicts total power. If the power supply exceeds the available capacity, UVC power will be maintained first. The speed of the fan remains unchanged, but the fan speed is reduced. Balance energy consumption and record wind farm degradation events.

[0048] Step S303: The multi-source cooperative control system 140 converts mathematical commands into physical field actions through low-level driving, constructing a three-dimensional turbulent field with time-varying characteristics. For the nonlinear characteristics of the brushless motor, a nonlinear PWM drive signal is generated. The system constructs a mapping model including dead-time compensation and load feedforward to calculate the real-time duty cycle. : ; in, To set the dead zone threshold; This is the aerodynamic load compensation coefficient, used to compensate for the nonlinear increase in air resistance under high-speed rotation. The dynamic scanning system of the airflow space vector coordinates with the deflection angle of the guide vanes. This causes it to perform periodic jitter: ; in, The scan range; The space sweep frequency (set to be lower than the fan pulsation frequency); This is the phase synchronization angle, used to set the spatiotemporal alignment between the maximum wind speed and a specific scanning angle. Closed-loop speed correction based on back EMF utilizes back EMF to detect the actual rotational speed. The duty cycle correction is calculated using an incremental PID algorithm. Output the final duty cycle This closed-loop mechanism eliminates the impact of battery voltage fluctuations or reverse flow on flow field stability. The fan abnormality monitoring and self-cleaning system monitors the operating current in real time. If the current is detected to continuously exceed the stall threshold... If the fan speed is abnormal, it is determined that foreign objects have been sucked in, triggering a reverse self-cleaning program: the fan is instructed to stop rotating forward and perform a short-term full-speed reverse rotation, using the reverse airflow to remove debris from the protective net, and then automatically resumes operation.

[0049] Step S400: Throughout the entire operation cycle, the load monitoring module 114 collects the total load power of the chassis drive and actuators in real time. The energy power system 110 dynamically schedules energy sources based on the ambient light irradiance and the state of charge (SOC) of the energy storage module 112: when there is sufficient sunlight, solar energy is used first; under heavy load or nighttime conditions, the fuel generator 111 is started and the power generation is adjusted according to the SOC stratification strategy to maintain a dynamic balance between load, power generation, and energy storage.

[0050] Step S401: Monitoring module 114 establishes a power sensing model with anti-interference and trend prediction capabilities to provide data support for dynamic energy allocation. Multi-channel electrical parameter acquisition and filtering: Load monitoring module 114 synchronously acquires the real-time current of each circuit. With bus voltage And perform timestamp alignment. For electromagnetic interference and load impacts in farmland environments, an exponentially weighted moving average (EWMA) algorithm is used to extract robust load characteristics, and the smoothed power values ​​are then analyzed. The calculation is as follows: ; in, A smoothing factor (ranging from 0.1 to 0.3) is used, determined based on the sampling frequency and load response time constant, to filter out transient spikes while preserving the true load trend. The load component decomposition and operating condition identification module breaks down the total power and calculates the actual drive power. (Total power minus deterministic loads such as UVC and wind turbines), and compared with the theoretical load power based on the vehicle dynamics model. Compare: ; in, The speed of travel; The slope angle; This is the rolling resistance coefficient. If Greater than Furthermore, if the load continues to exceed the preset window, a high-resistance condition label is generated, indicating potential muddy slippage or mechanical jamming. The power demand trend forecasting incorporates a feedforward approach from model predictive control, predicting the future window based on terrain data along the path. Peak power demand within : ; in, The elevation difference of the terrain ahead is used to calculate the potential energy power increment; The standard deviation of load power characterizes the degree of fluctuation. This is the safety margin factor. The power supply capacity safety boundary check, combined with the state of health (SOH) of energy storage module 112, calculates the current discharge rate. : ; like Exceeding the maximum allowable multiplier If the predicted power exceeds the system limit, an overload degradation request will be triggered immediately to proactively reduce non-critical loads.

[0051] Step S402: The energy and power system 110 constructs a multi-objective decision-making mechanism based on an efficiency and lifespan coupling model to dynamically adjust the energy flow direction of photovoltaic, fuel oil, and energy storage. The photovoltaic energy priority capture system monitors the photovoltaic array in real time through the MPPT controller and calculates the effective photovoltaic power by combining real-time illuminance and the power temperature coefficient of the photovoltaic modules. Priority is given to feeding into the bus. If the net load demand after deducting the photovoltaic contribution is less than zero, the system automatically switches to pure photovoltaic power supply and utilizes the surplus for charging. The energy management strategy based on SOC zoning introduces continuous gain control based on the S-shaped logic function. It dynamically adjusts the involvement level of the fuel generator according to the SOC of the energy storage module and calculates the power generation correction factor. : ; in, Maintain a threshold for the target; This is the charging intensity gain coefficient; This represents the regression slope. When the SOC is below the target value, Rapidly increase to promote rapid recharging; smoothly decrease as it approaches the target value to prevent overcharging. Optimal efficiency optimization at the fuel generator's operating point ensures the internal combustion engine operates within the optimal fuel consumption rate (BSFC) range. The system calculates the target output power based on the engine's universal characteristics. : ; in, and These represent the lower and upper limits of the generator's high-efficiency operating range power; This is the rated charging power. Anti-jitter start-stop control is employed to avoid frequent generator starts and stops caused by load fluctuations. The system incorporates asymmetric hysteresis judgment logic to determine the generation status. : ; Simultaneously set minimum runtime constraints This forces the generator to complete a full heat engine cycle.

[0052] Step S403: When charging the energy storage module 112 is required, a continuous flow control strategy based on multiple constraint boundaries is executed, instead of the traditional two-stage hard switching. The battery dynamic internal resistance and health correction system uses the voltage response at the moment of charging start-up or load change to identify the equivalent series internal resistance online. If the internal resistance is significantly higher than the calibrated value, it is determined that the state of equilibrium (SOH) has decreased, and the upper limit of the subsequent charging current is actively reduced. The environmental adaptability current boundary is set to prevent low-temperature lithium plating, and a temperature-current coupling constraint is constructed. The temperature decay coefficient is calculated using a nonlinear S-shaped function. This allows for the determination of the maximum safe charging current. : ; ; This logic establishes a safety boundary of strict control at low temperatures, relaxed control at room temperatures, and complete cutoff at high temperatures. The three-stage tiered charging strategy employs a multi-constraint parallel minimum value selection mechanism to calculate the target charging current command. This enables automatic and smooth transition between constant current, constant power, and constant voltage modes. ; The first term is physical security constraint; the second term is source-end power constraint; the third term utilizes... (Gain in the constant voltage region) Simulates negative feedback characteristics, in voltage approximation The current decays exponentially to prevent overcharging. Pulse elimination and end-of-life maintenance: When the SOC is high and the battery is shut down, short reverse discharge pulses (polarization elimination program) are periodically executed to disrupt the double-layer structure on the electrode surface, eliminate concentration polarization, and maintain battery consistency.

[0053] Step S500: During and after the operation, the remote operation and maintenance management module 144 continuously monitors the status of core components such as the luminous efficiency of UVC lamp beads, and generates an early warning if the status falls below the threshold. The dual-mode communication module 143 automatically selects and switches between 4G / 5G or LoRa links based on the on-site signal quality to upload the operation data. The cloud updates the land defect database accordingly, completing the operation loop and decision optimization.

[0054] Step S501: The remote operation and maintenance management module 144 constructs an online health (SOH) assessment model based on electrothermal-optical multi-physics field coupling to distinguish between device aging and external contamination. Quasi-steady-state synchronization and cleaning of multi-dimensional time-series data are performed to eliminate data phase deviation, and a current variance threshold is set. A quasi-steady state is determined to have been entered if and only if the variance of the driving current at N consecutive sampling points is less than this threshold, and the observation vector is then truncated. .in, The substrate thermodynamic temperature, The monitored relative irradiance is used. A moving average filter to remove extreme values ​​is employed to process the raw data. The intrinsic photoelectric efficiency algorithm based on thermal quenching compensation introduces a bandgap contraction theoretical model to eliminate the thermal quenching effect caused by high temperatures and calculates the intrinsic photoelectric conversion efficiency normalized to the standard temperature. : ; in, This is the thermal quenching coefficient (characterizing the sensitivity of luminous flux to temperature decay). This is the standard calibration temperature. This formula compensates for heat loss in reverse, restoring the true photoelectric conversion capability of the device. Electrical impedance feature extraction and trend tracking are used to simultaneously calculate the equivalent dynamic resistance. (Based on Ohm's law), and calculate its value relative to the initial calibration value. resistivity : ; Multidimensional health fusion and hierarchical decision-making based on smoothing processing and Implement hierarchical decision-making: Optical path pollution: Significant decline but Normal. The lens is determined to be contaminated, and a cleaning command is generated.

[0055] Device aging: Continued decline and If the increase is irreversible, it is determined to be physical aging of the chip, and a replacement warning is generated.

[0056] Circuit fault: If a step change occurs, it is determined to be an open circuit or short circuit. The power supply should be immediately cut off and an alarm should be triggered.

[0057] Step S502: The dual-mode communication module 143 incorporates a dynamic routing algorithm based on link quality service demand matching to establish a multi-dimensional evaluation model between broadband cellular (4G / 5G) and narrowband IoT (LoRa). Heterogeneous normalization of channel state information utilizes a logistic function to map heterogeneous physical parameters to a normalized score. Broadband link quality score. : ; Narrowband Link (LoRa) Quality Score Introduce bit error rate (PER) as a linear penalty: ; Feature vector extraction of business data stream involves performing deep packet inspection (DPI) on data packets to construct feature vectors. , representing payload volume, service priority, and maximum tolerable delay, respectively. Link arbitration based on the energy efficiency delay cost function calculates the transmission cost function of each link. The goal is to find the path with the minimum total cost. (Broadband cost) : ; Narrowband Cost (Including physical load limit constraints): ; in, For latency-to-energy conversion coefficients. To prevent frequent handovers, anti-jitter hysteresis handover execution introduces hysteresis comparison logic. Handover occurs if and only if the alternative link... The cost is significantly lower than the current link. Switching is performed on time: ; in, To switch the hysteresis threshold.

[0058] Step S503: In this embodiment, after the crop fungal disease control robot 100 completes its work on the designated plot and triggers a return command, the remote operation and maintenance management module 144 enters the final stage. This step aims to utilize the component health characteristics obtained in step S501 and the operational status transmitted in step S502 to perform dual updates on the entire life cycle of the component and the evolution of diseases in the plot.

[0059] After the operation is completed, the autonomous mobile chassis invokes the positioning and navigation module 123 to plan the optimal path from the current location to the docking point (GNSS for field environments, SLAM for greenhouse environments). During the return journey, the dual-mode communication module 143 uploads key indicators, structured and processed by the multi-source collaborative control system 140, to the cloud, including: Operational topology heatmap: Marks the coordinates of actual high-risk disease-infected areas visually identified during operations, used to update the historical disease distribution layer of the plot; Energy efficiency distribution spectrum: Records the actual fuel consumption rate and battery discharge curve under different terrain slopes, which are used to correct the load prediction model of the energy power system 110.

[0060] Based on cumulative stress-based component life prediction (RUL), given the requirement in the handover document that the component life prediction error be ≤ ±10%, the remote operation and maintenance management module 144 no longer uses the simple linear depreciation method. The intrinsic photoelectric conversion efficiency calculated in system call step S501 is used instead. Based on historical trends and the accumulated thermal stress from this operation, the lifespan loss factors of the UVC module and energy storage module are calculated. : ; in, For sampling current, The real-time junction temperature obtained in step S501 The aging rate constant is The activation energy is determined by the formula, which is based on the Arrhenius model and physically quantifies the nonlinear accelerated aging effect of high temperature and high load on device lifespan. When the remaining lifetime (RUL) calculated from accumulated losses is lower than the warning threshold (e.g., 15%), the system automatically generates a spare parts replacement work order. The cloud platform receives the uploaded data and iteratively corrects the disease risk decision model for the site after receiving the uploaded data. The system compares the predicted disease level before the operation with the actual disease hotspots identified during the operation. If the actual disease level in a certain area is found... Significantly higher than the predicted level If so, the initial risk weight for the next round of operations on that plot of land will be adjusted. : ; in, To correct the benefit coefficient. Through this step, when the robot enters the site again, the multi-source collaborative control system 140 will automatically load the updated strategy (such as increasing the base UVC power threshold or increasing the lower limit of the wind turbine speed), thereby ensuring that the prevention and control effect continuously approaches the optimal solution as the frequency of operation increases.

Claims

1. A method for controlling fungal diseases in crops based on ultraviolet light, characterized in that, Robots used for the control of fungal diseases in crops include: The multi-source sensor fusion module collects environmental parameters and combines them with the location information from the positioning and navigation module and historical disease data to generate an input vector. The disease risk decision-making module calculates the disease risk level based on the input vector; when the disease risk level reaches the threshold or the canopy three-dimensional scanning recognition unit identifies disease characteristics, the multi-source collaborative control system generates an operation instruction. In response to receiving the operation instruction, the canopy 3D scanning and recognition unit scans the crop canopy to obtain canopy height, canopy thickness and leaf density parameters; The multi-source collaborative control system calculates the target emission angle, target irradiation power, and extension stroke of the electric telescopic arm of the UVC intelligent irradiation module based on the canopy height, canopy thickness, and blade density parameters, and calculates the fan speed of the wind farm collaborative ventilation module based on the target irradiation power and the preset linkage ratio. The autonomous mobile chassis moves forward, and the electric telescopic arm delivers the UVC intelligent irradiation module to the working position. The irradiation posture is adjusted according to the target emission angle. The UVC intelligent irradiation module emits ultraviolet rays with the target irradiation power, and the wind field coordinated ventilation module generates airflow disturbance to the plant blades with the fan speed. During and after the operation, the remote operation and maintenance management module monitors the status and uploads data through the dual-mode communication module, updating the disease database and strategies.

2. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 1, characterized in that, The generation of the input vector and the calculation of the disease risk level based on the input vector include: The collected environmental parameters, crop status data, and historical disease data are spatiotemporally aligned and normalized to generate a normalized input vector. The outbreak rate of fungal diseases is calculated based on a multivariate logistic regression model, and the outbreak rate of fungal diseases is mapped to a discrete benchmark risk level. The canopy 3D scanning and recognition unit integrates a visual sensor and introduces a forced correction mechanism based on machine vision: when the visual sensor detects that the confidence level of the lesion phenotype exceeds the lower limit and the area of ​​the suspected lesion exceeds the minimum judgment threshold, the visual marker position is marked as valid. Risk arbitration is conducted based on the principle of prioritizing phenotypic evidence to calculate the final execution risk level: if the visual marker is valid, the final execution risk level is locked at no less than the preset minimum level threshold; if the visual marker is invalid, the larger value between the baseline risk level and the minimum level threshold is taken.

3. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 1, characterized in that, The data collected by the canopy 3D scanning and recognition unit is a canopy point cloud set; the calculation of the target emission angle of the UVC intelligent irradiation module based on the canopy height, canopy thickness, and leaf density parameters includes: The canopy point cloud set is divided into continuous differential slices along the direction of travel, and the average canopy height and average canopy thickness of each slice are calculated. The slice space is discretized into a three-dimensional voxel grid, the number of voxels occupied by the echo is counted, and the volumetric light resistance density index is calculated, which is used as a quantitative characterization of the blade density parameter. Based on the volumetric light resistance density index, a dynamic balance strategy is used to calculate the optimal light coupling distance of the UVC intelligent irradiation module: when the volumetric light resistance density index approaches 1, the distance is shortened to enhance penetration energy; when the volumetric light resistance density index approaches 0, the distance is increased to expand the light spot coverage area. Based on the principle of radiation centroid alignment, the target emission angle is calculated according to the vertical geometric center height of the canopy slice and the rotation axis height of the UVC intelligent irradiation module.

4. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 3, characterized in that, The target irradiation power is calculated using an exponential penetration compensation model: First, the target radiation dose is calculated using a graded gain model based on the real-time disease risk level. By combining the travel speed of the autonomous mobile chassis, the reference transmission power can be calculated. The reference emission power is nonlinearly compensated by introducing an exponential term with the volumetric light resistivity index as the independent variable to generate the target irradiation power; If the target irradiation power exceeds the hardware's rated power, the power will be truncated to the rated value, and a speed degradation request signal will be generated and fed back to the autonomous mobile chassis to maintain a constant cumulative radiation dose by reducing the travel speed.

5. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 3, characterized in that, The preset linkage ratio is a mapping relationship based on the dynamic adjustment of canopy impedance. The calculation of the fan speed of the wind field coordinated ventilation module includes: A layered aerodynamic impedance model was established to convert the volumetric light resistance density index of the canopy into the ideal penetration wind speed. Following the energy matching principle, the target fan speed is calculated, and the target fan speed is positively correlated with the target irradiation power. A sinusoidal disturbance is superimposed on the target wind turbine speed to generate a real-time dynamic command. The frequency of the sinusoidal disturbance is dynamically determined by the travel speed of the autonomous moving chassis to ensure that the blades complete multiple rotations during the period when the light window sweeps across. The operating current of the fan is monitored in real time. If the current continuously exceeds the stall threshold, the reverse self-cleaning program is triggered.

6. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 1, characterized in that, The environmental parameters include soil volumetric moisture content and local slope angle. The autonomous mobile chassis includes a dual-mode drive unit and a driving posture adaptive adjustment unit. The autonomous mobile chassis travels in the following steps: The dual-mode drive unit constructs a terrain traffic impedance index based on soil volumetric water content and local slope angle. The terrain passability impedance index is compared with the mode switching threshold: when the terrain passability impedance index is less than or equal to the mode switching threshold, the dual-mode drive unit is controlled to switch to wheel drive mode; when the terrain passability impedance index is greater than the mode switching threshold, the dual-mode drive unit is controlled to switch to track drive mode. The vehicle roll angle and pitch angle are monitored in real time by the driving posture adaptive adjustment unit, and the target vertical adjustment amount of the suspension support point is calculated. A semi-active ceiling damping control strategy is adopted, which adjusts the suspension damping coefficient in real time according to the vehicle's vertical absolute speed and the wheel's vertical speed to keep the UVC intelligent irradiation module's irradiation reference plane in a horizontal state.

7. The method for controlling fungal diseases in crops based on ultraviolet light according to claim 1, characterized in that, The updated disease database and strategy include: An online health assessment model based on electrothermal-optical multi-physics field coupling was constructed to calculate the intrinsic photoelectric conversion efficiency of the light-emitting unit in the UVC smart irradiation module normalized to the standard temperature, and to determine the aging status of the device by combining the equivalent dynamic resistance drift rate. Based on the Arrhenius model, the lifetime loss factor is calculated by combining real-time junction temperature and sampled current to predict the remaining lifetime of core components. Based on the remaining lifetime and device aging status, the power compensation strategy for subsequent operations is dynamically adjusted. After the operation is completed, the predicted disease level before the operation is compared with the actual disease hotspots identified during the operation. If the actual disease level is found to be significantly higher than the predicted level, the initial risk weight of the next round of operation for the plot is adjusted to correct the disease risk decision model.

8. A device for controlling fungal diseases in crops based on ultraviolet light, characterized in that, The method for controlling crop fungal diseases based on ultraviolet light, as described in any one of claims 1-7, comprises: A crop fungal disease control robot (100) includes an energy and power system (110), an autonomous mobile chassis (120), an ultraviolet irradiation execution system (130), and a multi-source collaborative control system (140). The energy and power system (110) provides power support for the movement and operation of the crop fungal disease control robot (100); The autonomous mobile chassis (120) carries the energy and power system (110), the ultraviolet irradiation execution system (130), and the multi-source collaborative control system (140). The ultraviolet irradiation execution system (130) is located above the autonomous mobile chassis (120) and is used for physical control of crop canopy. The multi-source collaborative control system (140) is communicatively connected to the energy power system (110), the autonomous mobile chassis (120), and the ultraviolet irradiation execution system (130), respectively.

9. The ultraviolet-based crop fungal disease control device according to claim 8, characterized in that, The energy power system (110) includes a fuel generator (111), an energy storage module (112), a solar charging module (113), and a load monitoring module (114); the fuel generator (111) and the energy storage module (112) are electrically connected to form a hybrid power supply circuit; the load monitoring module (114) is used to collect real-time power demand data of the whole machine and the state of charge data of the energy storage module (112); The autonomous mobile chassis (120) includes a dual-mode drive unit (121), a driving posture adaptive adjustment unit (122), a positioning and navigation module (123), and a road condition recognition module (124). The dual-mode drive unit (121) includes an all-wheel drive mechanism and a track drive mechanism, which are used to switch the drive mode according to terrain parameters. The driving posture adaptive adjustment unit (122) is connected to the suspension system and maintains the body level by adjusting the damping. The positioning and navigation module (123) is used to acquire position information. The ultraviolet irradiation execution system (130) includes a canopy three-dimensional scanning and recognition unit (131), a UVC intelligent irradiation module (132), an electric telescopic arm (133), and a wind field coordinated ventilation module (134). The canopy three-dimensional scanning and recognition unit (131) is used to collect data on the height, thickness, and leaf density of the crop canopy. The electric telescopic arm (133) is connected to the UVC intelligent irradiation module (132). The wind field coordinated ventilation module (134) is located on the side of the UVC intelligent irradiation module (132). The storage and folding mechanism includes an electric guide rail and a semi-cylindrical cavity located on the machine body. In the non-operating state, the lamp array of the UVC intelligent irradiation module (132) is driven by the electric guide rail to retract to the inside of the machine body and folds and attaches to the inner wall of the semi-cylindrical cavity so that there are no exposed lamps after it is fully stored.

10. The ultraviolet-based crop fungal disease control device according to claim 8, characterized in that, The multi-source collaborative control system (140) includes a multi-source sensor fusion module (141), a disease risk decision-making module (142), a dual-mode communication module (143), and a remote operation and maintenance management module (144). The multi-source sensor fusion module (141) is used to collect environmental data, crop status data and historical disease data; The disease risk decision module (142) is used to calculate the disease risk level based on the input vector; The remote operation and maintenance management module (144) monitors the operating status of each module in real time through the dual-mode communication module (143); The dual-mode communication module (143) has a built-in dynamic routing algorithm based on link quality and service demand matching. Based on broadband link quality score, narrowband link quality score and feature vector of service data flow, it calculates the transmission cost function of each link and switches between 4G / 5G broadband network and LoRa narrowband network according to the minimum transmission cost function.

Images