An intelligent spraying control system for a plant protection unmanned aerial vehicle

Abstract

The application discloses a kind of intelligent spraying control systems of plant protection unmanned plane, and it relates to the field of intelligent control technology.The kind of intelligent spraying control systems of plant protection unmanned plane, including data acquisition and preprocessing module, for collecting unmanned plane operation multi-source data and realizing preprocessing operation;Path planning and optimization module, for carrying out path comprehensive cost evaluation by trajectory coordinates, environmental sensing and operation condition data;Boundary area dynamic compensation module, for carrying out spraying disturbance analysis by flight state, wind speed disturbance and topographic feature data;Response lag compensation module, for carrying out execution lag analysis by timing characteristics, flow state and actuator data;Feedback control and optimization module, for iteratively updating control parameters.Solve the problem that plant protection unmanned plane is not uniform in boundary area spraying, flight path optimization is insufficient and spraying precision is reduced due to response lag of spraying equipment.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to an intelligent spraying control system for agricultural drones. Background Technology

[0002] With the development of agricultural automation technology, plant protection drones are increasingly widely used in agricultural production. As a highly efficient and flexible agricultural aerial spraying equipment, plant protection drones have been widely applied in modern agricultural production. Their advantages, such as high operating efficiency, strong terrain adaptability, and precise spraying, have significantly improved the timeliness and effectiveness of pest and disease control.

[0003] For example, invention patent CN116602286A discloses a control system and method for spraying agricultural drones, relating to the field of drone technology, to alleviate the adverse effects of insufficient flow meter accuracy on agricultural drone operations. The system includes a main control unit, a weighing unit, and a spraying unit. The weighing unit is used to acquire the mass of the pesticide solution, and the spraying unit is used to control the spraying flow rate. The main control unit is used to determine a pesticide flow rate prediction model based on historical spraying data. It determines a first spraying area based on the user-set pesticide dosage per acre and initial pesticide mass. It determines a second spraying area based on the pesticide dosage per acre, current pesticide mass, and the area already sprayed. The first and second spraying areas are input into the pesticide flow rate prediction model to determine the pesticide flow rate during the prediction period. The spraying control signal is determined based on the pesticide flow rate during the prediction period to improve the uniformity of pesticide spraying.

[0004] For example, the invention patent with announcement number CN118252136B discloses a method, device, equipment, and medium for controlling the spraying of agricultural drones. The method includes: acquiring the pose transformation data of each trunk of the fruit tree canopy from the pose sensor within a preset time range, various flight parameters of the agricultural drone, and fruit tree canopy image data from the visual sensor in the agricultural drone; determining the amount of droplet deposition in the fruit tree canopy under each flight parameter; constructing an agricultural drone spraying control model based on the amount of droplet deposition and the canopy disturbance prediction model; inputting the various flight parameters of the agricultural drone into the pre-trained agricultural drone spraying control model to determine the fruit tree canopy disturbance state corresponding to the flight parameters and the amount of droplet deposition under the fruit tree canopy disturbance state, so as to complete the control of the agricultural drone spraying, which can greatly avoid the huge impact of the wind field generated by the drone rotor on the deposition effect of pesticide droplets.

[0005] Therefore, in order to address the above problems, there is an urgent need for an intelligent spraying control system for agricultural drones. Summary of the Invention

[0006] Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an intelligent spraying control system for agricultural drones, which solves the problems of uneven spraying in boundary areas, insufficient flight path optimization, and decreased spraying accuracy due to lag in the response of spraying equipment.

[0007] Technical solution To achieve the above objectives, the present invention provides the following technical solution: an intelligent spraying control system for agricultural drones, comprising a data acquisition and preprocessing module for acquiring multi-source data from drone operations, preprocessing the multi-source data, storing it, and constructing an operation control database; a path planning and optimization module for conducting comprehensive path cost assessment using trajectory coordinates, environmental sensing data, and operational condition data, and performing adaptive optimization and smoothing control operations on the flight trajectory based on the cost assessment results; a boundary area dynamic compensation module for conducting spraying disturbance analysis using flight status, wind speed disturbance, and terrain feature data, and performing spraying parameter compensation operations on the boundary area based on the disturbance analysis results; a response lag compensation module for conducting execution lag analysis using time-series characteristics, flow rate status, and actuator data, and performing spraying output timing correction and abnormal section control operations based on the lag analysis results; and a feedback control and optimization module for integrating multi-module operating data and historical deviation samples to conduct intelligent optimization learning and iteratively update control parameters.

[0008] Furthermore, the specific steps for collecting multi-source data from UAV operations are as follows: Collecting multi-source data from UAV operations: Real-time acquisition of flight trajectory coordinates via the onboard positioning module and calculation of the difference in the trajectory's horizontal coordinates to obtain the horizontal distance of the activity path; real-time acquisition of wind speed data via environmental sensors; acquisition of flight altitude, flight speed changes, nozzle opening adjustment time, path planning start time, and path planning end time via the flight control system; real-time acquisition of farmland prescription maps and spray coverage via image sensors; real-time acquisition of flight angle changes via inertial measurement unit; real-time monitoring of the stable spray flow rate via flow monitoring sensor, synchronously recording the stable spray flow rate moment when the flow rate reaches a stable threshold; acquisition of standard spray flow rate via equipment calibration parameters, and real-time recording of the actuator's nozzle opening and pump pressure status data.

[0009] Further, the specific steps for storing and constructing an operation control database after preprocessing multi-source data of UAV operations are as follows: Each sensor record is synchronously written with a UAV identifier, acquisition time, task batch number, and geographic coordinate identifier; time alignment and window aggregation of multi-source data streams are performed based on the acquisition time; sliding window denoising, outlier removal, and missing value interpolation are performed on the multi-source data of UAV operations, integrating multi-source records within the same observation time window into operation observation segments, and assigning a globally unique segment sequence number to each segment; the UAV identifier, task batch number, and operation area identifier are uniformly encoded and standardized using a distribution standardization algorithm to achieve unified format of the identifier data; the multi-source data of UAV operations is normalized and scaled to a uniform numerical range using a minimum-maximum normalization algorithm; the standardized and normalized multi-source data of UAV operations and operation observation segments are appended with corresponding segment sequence numbers, UAV identifiers, and acquisition time window markers, stored, and the operation control database is constructed.

[0010] Furthermore, the specific steps for evaluating the comprehensive cost of a path using trajectory coordinates, environmental sensing data, and operational condition data are as follows: Obtain the horizontal distance of the path at the current location and perform an absolute value calculation to obtain the absolute value of the horizontal distance; obtain the wind speed at the current location and perform an absolute value calculation to obtain the absolute value of the wind speed; obtain the flight altitude at the current location and perform an absolute value calculation to obtain the absolute value of the flight altitude; obtain the original value of the spray coverage and perform an absolute value calculation to obtain the absolute value of the spray coverage; sum the above four absolute values ​​to obtain the instantaneous comprehensive cost; perform a definite integral calculation on the instantaneous comprehensive cost from the start time to the end time of path planning to obtain the optimized path value.

[0011] Furthermore, the specific steps for performing adaptive optimization and smoothing control of the flight trajectory based on the cost assessment results are as follows: By comparing the path optimization value with the optimization threshold in real time, when the path optimization value is less than the optimization threshold, the existing flight strategy is maintained, and operations are carried out continuously and stably; when the path optimization value is greater than or equal to the optimization threshold, the first-order partial derivatives of path length, wind speed interference, flight altitude, and spray coverage are calculated and sorted in descending order to obtain the path optimization impact queue; the dominant influencing factors are identified based on the path optimization impact queue, and corresponding treatment plans are implemented: when the dominant factor is path length, A... The algorithm optimizes path planning and smooths the flight trajectory using cubic B-spline curves, reducing turning maneuvers and reciprocating flight. When wind speed is the dominant factor, a dynamic wind field model is constructed based on real-time wind field data collected by the UAV's onboard environmental sensors. The flight trajectory is adjusted in conjunction with the wind field model to reduce the impact of wind on trajectory stability. When flight altitude is the dominant factor, the trajectory elevation change is optimized based on terrain undulations, and the ascent and descent amplitudes are smoothly adjusted to eliminate the impact of abrupt altitude changes on trajectory smoothness. When spray coverage is the dominant factor, a grid partitioning algorithm is used to divide the work area into m×m grids and calculate the trajectory coverage frequency of each grid. For areas with a coverage frequency lower than the frequency threshold, a greedy algorithm is used to supplement the trajectory path, and the algorithm iterates every k seconds until the calculated path optimization value falls back to within the optimization threshold, thus achieving dynamic optimization and stable control of the flight trajectory.

[0012] Furthermore, the specific steps for conducting spraying disturbance analysis using flight status, wind speed disturbance, and terrain feature data are as follows: Obtain the standard spraying flow rate value, flight speed variation, flight angle variation, and wind speed; calculate the product of the flight speed variation and the flight speed compensation coefficient to obtain the speed compensation term; calculate the product of the flight angle variation and the flight angle compensation coefficient to obtain the angle compensation term; calculate the product of the wind speed and the wind speed compensation coefficient to obtain the wind speed compensation term; calculate the sum of the speed compensation term, angle compensation term, and wind speed compensation term, and add one to obtain the comprehensive compensation term; multiply the standard spraying flow rate value by the comprehensive compensation term to obtain the spraying compensation value.

[0013] Furthermore, the specific steps for implementing spray parameter compensation operations in the boundary area based on the disturbance analysis results are as follows: By comparing the spray compensation value with the spray threshold in real time, when the spray compensation value is less than the spray threshold, the flight attitude is stable, maintaining the existing spray output and travel rhythm; when the spray compensation value is greater than or equal to the spray threshold, the plot boundary is obtained by combining UAV onboard GNSS positioning with farmland prescription maps, and a boundary buffer zone is generated with a distance of B meters from the boundary as the buffer width; the main sources of disturbance are determined by sorting the speed compensation item, the ratio of the speed compensation item and the wind speed compensation item to the comprehensive compensation item in descending order, and addressing the disturbance caused by changes in flight speed. The system employs a PID control algorithm to adjust the flight speed and reduce speed fluctuations. For disturbances caused by changes in flight angle, a fuzzy control algorithm is used to optimize the attitude adjustment process, reducing the amplitude and frequency of attitude changes and maintaining a stable spraying output. For disturbances caused by wind speed, real-time wind speed data collected by onboard environmental sensors is input, and the correlation between wind speed and spraying output is fitted using the least squares method. The output variable is the spraying flow rate adjustment, and the spraying output scale is adjusted synchronously to offset the spraying offset caused by the external wind field until the calculated spraying compensation value falls back within the spraying threshold. The flight control record is then archived in the operation control database.

[0014] Further, the specific steps for conducting execution lag analysis based on time-series characteristics, flow status, and actuator data are as follows: Obtain the spray compensation value, the stable spray flow time, and the nozzle opening adjustment time; calculate the difference between the stable spray flow time and the nozzle opening adjustment time to obtain the compensation time; multiply the compensation time by the lag response attenuation coefficient to obtain the time attenuation product term; take the negative value of the time attenuation product term to obtain the negative time attenuation term; perform natural exponential calculation on the negative time attenuation term to obtain the exponential attenuation term; calculate and subtract the exponential attenuation term to obtain the lag response correction term; obtain the lag compensation coefficient, which is generated iteratively by the feedback control and optimization module based on historical deviation samples and stored in the operation control database; calculate the product of the lag compensation coefficient, the spray compensation value, and the lag response correction term to obtain the lag compensation value.

[0015] Furthermore, the specific steps for performing spray output timing correction and abnormal section control operations based on the lag analysis results are as follows: By comparing the lag compensation value and the lag threshold in real time, when the lag compensation value is less than the lag threshold, the spray response is stable, the output and control commands remain synchronized, and the current spraying state continues to operate; when the lag compensation value is greater than or equal to the lag threshold, a first-order lead control algorithm with a time constant of τ and a gain of l is used to correct the spray output timing, shorten the response delay of the nozzle and pump body, and eliminate the time deviation between the spraying action and flight control. If the lag compensation value calculated after n consecutive rounds of treatment is still greater than or equal to the lag threshold, the current operating section is marked as an abnormal response delay section and transferred to the deviation sample queue. At the same time, the lag compensation coefficient is frozen for adaptive updating, and the current lag compensation coefficient value in the operation control database remains unchanged.

[0016] Furthermore, the specific steps for integrating multi-module operational data and historical deviation samples to conduct intelligent optimization learning and iteratively update control parameters are as follows: Flight trajectory coordinates, wind speed, and spray flow rate data are used as basic inputs. Based on flight control records and environmental sensor records, cumulative distribution deviation records, deviation region grid markers, and deviation dominant factor ranking records are generated in the operation control database. These are combined with path optimization values, spray compensation values, hysteresis compensation values, and historical deviation samples to form a deviation analysis input set. A Bayesian optimization algorithm is used for learning and training to fit the nonlinear mapping relationship between cumulative distribution deviation within the deviation region and flight, environmental, and control parameters, ensuring that the predicted deviation under various effective operational conditions remains consistent with actual observations. This algorithm aims to minimize the deviation hot zone area, spray coverage variance, and spray drift outside the boundary as comprehensive optimization objectives. It uses a set of compensation coefficients, including speed, angle, wind speed, and hysteresis compensation coefficients, and path planning weights, including path length, wind speed interference, flight altitude, and spray coverage weights, as decision variables. Simultaneously, it limits pump pressure, flow rate, flight speed, and altitude to within the equipment's rated safe range, and defines the area outside the boundary as a no-spray zone. The algorithm uses the deviation hot zone to drive the optimization of compensation coefficients and path weights to improve spray uniformity, outputs the optimized compensation coefficient set, and updates the path cost weights, spray compensation coefficients, and hysteresis compensation coefficients. The optimized parameters are written into the operation control database to achieve closed-loop iterative updates of the control parameters of each module.

[0017] Beneficial effects The present invention has the following beneficial effects: (1) This invention constructs an operation control database by collecting data from multiple sources, aligning time, denoising and interpolating, and standardizing and normalizing the data. This ensures the integrity, standardization and analyzability of UAV operation data, thereby achieving the effect of precise control of operation data in all dimensions. It effectively solves the problems of messy, inconsistent dimensions and large errors in UAV operation data in the prior art, which lead to the distortion of subsequent analysis results.

[0018] (2) This invention improves the smoothness and adaptability of flight trajectory by constructing a comprehensive cost evaluation system based on path horizontal distance, wind speed, flight altitude and spray coverage, and by implementing differentiated trajectory optimization strategies for the dominant influencing factors, thereby achieving the effect of dynamic optimization and stable control of flight trajectory. It effectively solves the problems of path planning ignoring the coupling influence of multiple factors, uneven trajectory leading to low operation efficiency and uneven coverage in the prior art.

[0019] (3) This invention, by combining GNSS positioning and farmland prescription map to delineate boundary buffer zones, by quantifying the proportion of each disturbance factor to identify the main disturbance source and implementing targeted compensation strategies, accurately adjusts the spraying output parameters, thereby achieving the effect of dynamic adaptation of spraying parameters in the boundary area, effectively solving the problem of uneven coverage and drift outside the boundary caused by wind speed, velocity and angle disturbances in the existing technology.

[0020] (4) This invention integrates multi-module operation data and historical deviation samples, fits the deviation mapping relationship through Bayesian optimization algorithm and iteratively updates the compensation coefficient and path weight, thereby realizing closed-loop optimization of control parameters and achieving intelligent self-adaptation of operation parameters. This effectively solves the problem of fixed control parameters and inability to adapt to complex operation scenarios in the prior art.

[0021] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0022] Figure 1 This is a structural diagram of an intelligent spraying control system for a plant protection drone according to the present invention; Figure 2 This is a flowchart illustrating the process architecture for comprehensive cost evaluation and multi-factor collaborative optimization of the path in this invention. Figure 3 This is a scatter plot showing the relationship between the dynamic response of the spray flow rate and the velocity variation of the present invention. Figure 4 This is a comparison diagram of the exponential lag compensation and the pure lag step response of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figures 1-4This invention provides a technical solution: an intelligent spraying control system for agricultural drones, comprising a data acquisition and preprocessing module for acquiring multi-source data of drone operations, preprocessing the multi-source data, storing it, and constructing an operation control database; a path planning and optimization module for performing comprehensive path cost assessment based on trajectory coordinates, environmental sensing data, and operational condition data, and performing adaptive optimization and smoothing control operations on the flight trajectory based on the cost assessment results; a boundary area dynamic compensation module for performing spraying disturbance analysis based on flight status, wind speed disturbance, and terrain feature data, and performing spraying parameter compensation operations on the boundary area based on the disturbance analysis results; a response lag compensation module for performing execution lag analysis based on time sequence characteristics, flow status, and actuator data, and performing spraying output timing correction and abnormal section control operations based on the lag analysis results; and a feedback control and optimization module for integrating multi-module operating data and historical deviation samples to perform intelligent optimization learning and iteratively updating control parameters.

[0025] Specifically, the steps for collecting multi-source data on UAV operations are as follows: Collecting multi-source data on UAV operations: The latitude and longitude coordinates of the flight trajectory are acquired in real time via the onboard positioning module. After being converted to planar coordinates using Gauss-Kruger projection, the difference in the abscissa of adjacent trajectory points is calculated and accumulated within a time window to obtain the horizontal distance of the activity path; wind speed data is collected in real time via environmental sensors; flight altitude, flight speed changes, nozzle opening adjustment time, path planning start time, and path planning end time are acquired via the flight control system; farmland prescription maps are acquired in real time via image sensors, and the spraying coverage area is extracted from the images using semantic segmentation algorithms. The spray coverage is calculated as the ratio of the actual area covered by the deposited pesticide solution to the area of ​​the target region, with the output range normalized to between zero and one. The flight angle change is acquired in real time through an inertial measurement unit. The spray flow rate stability is monitored in real time through a flow monitoring sensor. When the flow rate fluctuation rate is less than 5% for two consecutive seconds, the flow rate is determined to have reached the stability threshold, and this stable moment is recorded as the spray flow rate stability moment. The standard spray flow rate is obtained by looking up a table using the nozzle flow rate-opening-pressure relationship curve obtained from the equipment's factory calibration, combined with the current nozzle opening and pump pressure status data, and the nozzle opening and pump pressure status data of the actuator are recorded in real time.

[0026] In this implementation plan, multi-source sensors are used to collect real-time data on the drone's flight trajectory, ambient wind speed, flight status, spray coverage, attitude changes, flow rate, and actuator data. This enables the synchronous acquisition and feature extraction of motion status, environmental disturbances, operational effects, and spray execution information during the operation. This provides comprehensive and accurate real-time data support for subsequent path optimization, dynamic compensation, lag correction, and feedback learning modules, ensuring that the intelligent control system can make accurate decisions based on complete on-site information.

[0027] Specifically, the steps for preprocessing and storing multi-source data from UAV operations and constructing an operation control database are as follows: Each sensor record is synchronously written with the UAV identifier, acquisition time, task batch number, and geographic coordinate identifier. Time alignment and window aggregation of the multi-source data streams are performed based on the acquisition time. Linear interpolation is used to synchronize sensor data from different sampling frequencies to a unified time reference, ensuring strict alignment of sensor data in the time dimension. Sliding window denoising, outlier removal, and missing value interpolation are performed on the multi-source data from UAV operations. The sliding window length is 5 sampling points. Median filtering is used for denoising. Outliers are identified and removed based on the 3σ principle. Missing values ​​are filled using linear interpolation of adjacent valid points. Multi-source records within the same observation time window are integrated into operation observation segments, and each segment is assigned a globally unique segment sequence number, consisting of a timestamp and a four-digit serial number. The distribution standardization algorithm uniformly encodes and standardizes the identification of UAVs, mission batch numbers, and operation areas, achieving a unified format for the identification data. The encoding rule converts each category of identification into an integer index with a value range of 1-9999, ensuring the uniqueness and consistency of all identifications in the database. The min-max normalization algorithm is used to normalize numerical features such as flight altitude, wind speed, flight speed variation, flight angle variation, and spray flow rate in the multi-source data of UAV operations, scaling them to a unified numerical range. This eliminates the influence of different physical dimensions on subsequent analysis, achieving dimensionless processing. The standardized and normalized multi-source data of UAV operations and operation observation segments are then appended with corresponding segment sequence numbers, UAV identifications, and acquisition time window markers for storage and construction of an operation control database. This provides a high-quality, uniformly formatted data foundation for subsequent path planning optimization, dynamic compensation, and feedback learning.

[0028] In this implementation scheme, the original heterogeneous sensor data is integrated into structured operation observation segments through time alignment, window aggregation, and quality cleaning of multi-source data, and a globally unique segment sequence number is assigned. Data format is unified by standardizing the identification information, and the influence of different physical dimensions is eliminated by using the minimum-maximum normalization algorithm to form a dimensionless standardized sensor sequence. Finally, the processed data, along with the corresponding segment sequence number, UAV identifier, and acquisition time window mark, are stored together to construct an operation control database, providing high-quality, unified-format real-time data support for subsequent path optimization, dynamic compensation, and feedback learning.

[0029] Specifically, the steps for comprehensive path cost assessment using trajectory coordinates, environmental sensing data, and operational condition data are as follows: First, obtain the horizontal distance of the path at the current location. This horizontal distance has been converted to a dimensionless value through min-max normalization in the data acquisition and preprocessing module. Then, perform an absolute value operation on this value to obtain the absolute value of the horizontal distance. Second, obtain the wind speed at the current location. This wind speed data has also been normalized and converted to a dimensionless value. Then, perform an absolute value operation on this value to obtain the absolute value of the wind speed. Third, obtain the flight altitude at the current location. This flight altitude data has been normalized and converted to a dimensionless value. The absolute value of the flight altitude is obtained by performing an absolute value operation on the flight altitude. The original value of the spray coverage is obtained, which is a dimensionless ratio within the range of 0-1. The absolute value of the spray coverage is then obtained by performing an absolute value operation on the spray coverage. The above four dimensionless absolute value terms are summed to obtain the instantaneous comprehensive cost value. The instantaneous comprehensive cost value is then integrally calculated from the start time of path planning to the end time of path planning to obtain the path optimization value. This path optimization value is a dimensionless comprehensive evaluation index used to quantitatively evaluate the comprehensive performance of the flight path under multiple dimensions such as length, environmental disturbance, and operational effectiveness.

[0030] The specific calculation method for the path optimization value is as follows: In the formula, This represents the optimized path value, reflecting the combined impact of flight path length and environmental factors in path planning; It represents the horizontal distance between waypoints, reflects the path length in path planning, and is obtained through path coordinate calculation; This indicates the wind speed at the current location, which affects spraying accuracy. This indicates the current flight altitude, which affects the uniformity of spraying. This indicates the coverage area, taking into account crop distribution. Indicates the start time of path planning; This indicates the end time of path planning, reflecting the time it takes for the task to be completed.

[0031] In this implementation plan, by summing the four dimensionless absolute values ​​of the normalized path horizontal distance, wind speed, flight altitude, and spray coverage, an instantaneous comprehensive cost value that can comprehensively reflect the flight length, environmental disturbances, and operational effects is constructed. Then, by performing definite integral calculation on the instantaneous comprehensive cost value within the time range of the path planning start and end times, a dimensionless path optimization value is obtained, realizing a multi-dimensional quantitative evaluation of the overall performance of the entire flight path, and providing a unified decision-making basis for the dynamic optimization and stable control of subsequent flight strategies.

[0032] Specifically, the steps for performing adaptive optimization and smoothing control of the flight trajectory based on the cost assessment results are as follows: By comparing the optimized path value with the optimized threshold in real time, such as... Figure 2 This is a flowchart illustrating the comprehensive cost assessment and multi-factor collaborative optimization process architecture of this embodiment. When the path optimization value is less than the optimization threshold, the existing flight strategy is maintained, and operations continue stably. When the path optimization value is greater than or equal to the optimization threshold, first-order partial derivatives are calculated for path length, wind speed interference, flight altitude, and spray coverage, respectively. This involves calculating the partial derivatives of the path optimization value with respect to the absolute values ​​of the path horizontal distance, wind speed, flight altitude, and spray coverage, respectively, to obtain the instantaneous change contribution rate of each factor under the current state. The partial derivative values ​​are then sorted in descending order to obtain the path optimization impact queue. Based on the path optimization impact queue, the dominant influencing factor is identified, and the corresponding handling plan is executed: when the dominant factor is path length, A... The algorithm optimizes path planning and smooths the flight trajectory using cubic B-spline curves, reducing turning maneuvers and reciprocating flight. When wind speed is the dominant factor, a dynamic wind field model is constructed using Kalman filtering based on real-time wind field data collected by the UAV's onboard environmental sensors. This model is then used to adjust the flight trajectory, reducing the impact of wind on trajectory stability. When flight altitude is the dominant factor, the trajectory elevation change is optimized based on terrain undulations, and the ascent and descent amplitudes are smoothly adjusted to eliminate the impact of abrupt altitude changes on trajectory smoothness. When spray coverage is the dominant factor, a grid partitioning algorithm is used to divide the work area into m×m grids, where m is 2 meters. A frequency threshold of 2 is set according to crop type and operational requirements. The trajectory coverage frequency of each grid is calculated. For areas with a coverage frequency lower than the frequency threshold, a greedy algorithm is used to supplement the trajectory path. The algorithm iterates every k seconds until the calculated path optimization value falls back to within the optimization threshold, where k is 3 seconds, achieving dynamic optimization and stable control of the flight trajectory.

[0033] In this implementation plan, dynamic decision-making and adaptive optimization of the flight strategy are achieved by comparing the optimized path value with the optimization threshold in real time: when the optimized path value does not exceed the optimization threshold, the current flight strategy is maintained in a stable operation; when the optimized path value exceeds the optimization threshold, the dominant disturbance factor under the current operating condition is accurately identified by calculating the first-order partial derivatives of each influencing factor and sorting them in descending order, and corresponding optimization measures are triggered for different dominant factors, including using A... The algorithm combines cubic B-spline curves to optimize path smoothness, constructs a dynamic wind field model based on airborne wind field data to adjust the heading, optimizes trajectory elevation changes based on terrain undulations, and uses grid subdivision and greedy algorithms to supplement the trajectory in weak areas. By iterating and updating every 3 seconds until the path optimization value falls back to within the optimization threshold, it achieves smooth control and multi-objective collaborative optimization of the flight trajectory in complex dynamic environments, effectively improving the stability, uniformity, and environmental adaptability of the operation process.

[0034] Specifically, the steps for spraying disturbance analysis using flight status, wind speed disturbance, and terrain feature data are as follows: Obtain the standard spraying flow rate, flight speed variation, flight angle variation, and wind speed; the flight speed compensation coefficient is obtained through experimental data fitting, i.e., measuring the actual spraying flow rate variation corresponding to different flight speed variations under standard operating conditions, and using the least squares method to fit the coefficient, with a value range of 0.15~0.25; the product of the flight speed variation and the flight speed compensation coefficient is calculated to obtain the speed compensation term; the flight angle compensation coefficient is obtained through flight attitude disturbance experiments, and the effect of flight angle variation on spray distribution is analyzed. The influence of the system was investigated, and the angle compensation coefficient was obtained through systematic identification, with a value range of 0.008 to 0.015. The angle compensation term was obtained by multiplying the change in flight angle by the flight angle compensation coefficient. The wind speed compensation coefficient was obtained through a combination of wind tunnel experiments and field tests. Based on the spray offset data under different wind speed conditions, the wind speed compensation coefficient was obtained through regression analysis, with a value range of 0.04 to 0.08. The wind speed compensation term was obtained by multiplying the wind speed by the wind speed compensation coefficient. The comprehensive compensation term was obtained by summing the speed compensation term, angle compensation term, and wind speed compensation term. The spray compensation value was obtained by multiplying the standard spray flow rate value by the comprehensive compensation term.

[0035] The specific formula for the spray compensation value is as follows: ; In the formula, This indicates the spray compensation value, reflecting the adjustment of the spray flow rate; It indicates the standard spraying flow rate and reflects the baseline spraying volume; Indicates the change in flight speed; It represents the change in flight angle and reflects changes in flight attitude; Indicates wind speed; The flight speed compensation coefficient is used to adjust the effect of flight speed on spray flow rate. This is the flight angle compensation coefficient, used to adjust the effect of the flight angle on the spray flow rate. This is the wind speed compensation coefficient, used to adjust the effect of wind speed on spray flow rate.

[0036] Table 1 shows the spray disturbance analysis and compensation data in this embodiment. For the first sampling point, the standard spray flow rate is 2.00, the velocity compensation term is 0.0216, the angle compensation term is 0.0420, the wind speed compensation term is 0.1680, and the comprehensive compensation term is 1.2316, resulting in a calculated spray compensation value of 2.46. For the second sampling point, the standard spray flow rate is 2.00, the velocity compensation term is -0.0144, the angle compensation term is -0.0252, the wind speed compensation term is 0.1920, and the comprehensive compensation term is 1.1524, resulting in a calculated spray compensation value of 2.30. The standard spray flow rate for the third sampling point is 2.00, the velocity compensation is 0.0090, the angle compensation is 0.0216, the wind speed compensation is 0.2700, and the comprehensive compensation is 1.3006, resulting in a calculated spray compensation value of 2.60. The standard spray flow rate for the fourth sampling point is 2.00, the velocity compensation is -0.0270, the angle compensation is 0.0504, the wind speed compensation is 0.0900, and the comprehensive compensation is 1.1134, resulting in a calculated spray compensation value of 2.23.

[0037] Table 1. Spraying Disturbance Analysis and Compensation Data Table like Figure 3 The figure shows a scatter plot of the dynamic response of spray flow rate and velocity variation provided in the embodiments of this application. Combined with the data in Table 1... Figure 3 As can be seen, at the third sampling point, the change in flight speed is 0.05, and the spray compensation value is 2.60. The wind speed compensation term at this point, 0.2700, is the highest among the four groups, corresponding to a wind speed of 4.5, indicating that strong wind disturbance is the dominant factor in the jump in compensation value. At the fourth sampling point, the change in flight speed is -0.15, and the spray compensation value is 2.23. The speed compensation term at this point, -0.0270, has the largest negative contribution, and the wind speed compensation term, 0.0900, is relatively small, both contributing to the lowest compensation value among the four groups. The slope of the trend line reflects the strength of the positive influence of the change in flight speed on the spray compensation value, while the vertical dispersion of the scatter points reflects the combined effect of the angle and wind speed compensation terms. This verifies that the spray compensation value formula can effectively integrate multi-source disturbance factors, providing a quantitative basis for dynamic compensation in the boundary area.

[0038] In this implementation plan, by quantifying the dynamic impact of changes in flight speed, flight angle, and ambient wind speed on the spray flow rate, speed compensation, angle compensation, and wind speed compensation are calculated. The three compensation terms are summed and then multiplied by one to obtain a comprehensive compensation term, which is then multiplied by the standard spray flow rate value to obtain the spray compensation value. This achieves real-time response and quantitative compensation to flight state fluctuations and external environmental disturbances, ensuring that the spray flow rate can dynamically adapt to changes in current operating conditions and effectively improve the spray uniformity and control accuracy in boundary areas and under disturbance conditions.

[0039] Specifically, the steps for implementing spray parameter compensation in the boundary area based on the disturbance analysis results are as follows: By comparing the spray compensation value with the spray threshold in real time, when the spray compensation value is less than the spray threshold, the flight attitude remains stable, maintaining the existing spray output and travel rhythm; when the spray compensation value is greater than or equal to the spray threshold, the plot boundary is obtained through UAV-borne GNSS positioning combined with a farmland prescription map, and a boundary buffer zone is generated with a distance of B meters from the boundary as the buffer width, where B is 2 meters; the ratios of the speed compensation term, angle compensation term, wind speed compensation term, and comprehensive compensation term are calculated respectively, i.e., the ratios of the speed compensation term to the comprehensive compensation term, the angle compensation term to the comprehensive compensation term, and the wind speed compensation term to the comprehensive compensation term. These three ratios are then sorted in descending order to determine the main sources of disturbance. For disturbances caused by changes in flight speed, a PID control algorithm is used to adjust... For flight speed, the aim is to minimize the variance of speed fluctuations to reduce the range of speed fluctuations. For disturbances caused by changes in flight angle, a fuzzy control algorithm is used to optimize the attitude adjustment process, with an angle change rate of ≤5° / s as a constraint to reduce the amplitude and frequency of attitude changes and maintain a stable spraying output process. For disturbances caused by wind speed, wind speed data collected in real time by airborne environmental sensors is input, and the correlation between wind speed and spraying output is fitted using the least squares method. A first-order linear model is used as the fitting function, and the output variable is the spraying flow rate adjustment. The spraying output scale is adjusted simultaneously to offset the spraying offset caused by the external wind field. The above treatment scheme is iterated and executed every 3 seconds. When the spraying compensation value calculated for three consecutive sampling cycles is less than the spraying threshold, it is determined that it has fallen back to within the spraying threshold, the iterative optimization stops, and the flight control record is archived to the operation control database.

[0040] In this implementation plan, dynamic identification and adaptive compensation of spraying disturbances in the boundary area are achieved by comparing the spraying compensation value with the spraying threshold in real time: when the spraying compensation value is less than the spraying threshold, stable operation is maintained; when the spraying compensation value reaches or exceeds the spraying threshold, the plot boundary is first determined based on airborne GNSS positioning and farmland prescription map, and a 2-meter buffer zone is generated. Then, by calculating the ratio of speed compensation, angle compensation, wind speed compensation and comprehensive compensation and sorting them in descending order, the main sources of disturbance under the current working condition are accurately identified, and corresponding control strategies are triggered for different disturbance types. These include using PID control algorithm to adjust the travel speed to reduce the speed fluctuation range, using fuzzy control algorithm to optimize the attitude adjustment process to reduce the attitude change amplitude and frequency, and using least squares method to fit the correlation between wind speed and spraying output and output spraying flow adjustment to offset wind field offset. The above optimization is iterated every 3 seconds until the spraying compensation value falls back to within the spraying threshold for three consecutive sampling cycles, thus achieving accurate and stable control of the boundary area under multi-source disturbances.

[0041] Specifically, the steps for conducting execution lag analysis using time-series characteristics, flow status, and actuator data are as follows: Obtain the spray compensation value, the spray flow stabilization time, and the nozzle opening adjustment time. The nozzle opening adjustment time refers to the moment the flight control system sends the opening adjustment command to the lower-level actuator, which is recorded in real-time by the flight control system and used as the time-series starting point. The spray flow stabilization time is synchronously recorded by the flow monitoring sensor when the flow fluctuation rate is less than 5% within two consecutive seconds. The difference between the spray flow stabilization time and the nozzle opening adjustment time is calculated to obtain the compensation time. The compensation time is multiplied by the lag response attenuation coefficient to obtain the time attenuation product term. The lag response attenuation... The coefficients are obtained through equipment step response testing, i.e., measuring the response curve from a step change in nozzle opening to a stable flow rate, and obtaining the attenuation coefficient through exponential fitting, with a value range of 1.5~2.5s⁻¹; taking the negative of the time attenuation product term to obtain the negative time attenuation term; performing natural exponential operation on the negative time attenuation term to obtain the exponential attenuation term; calculating and subtracting the exponential attenuation term to obtain the hysteresis response correction term; obtaining the hysteresis compensation coefficient, which is generated iteratively by the feedback control and optimization module based on historical deviation samples and stored in the operation control database; calculating the product of the hysteresis compensation coefficient, the spraying compensation value, and the hysteresis response correction term to obtain the hysteresis compensation value.

[0042] The specific formula for calculating the lag compensation value is as follows: ; In the formula, This represents the lag compensation value, which compensates for the response delay of the spraying equipment to ensure that the spraying flow rate is synchronized with the flight control. This indicates the spray flow compensation value, reflecting the amount of adjustment to the spray flow rate; The lag compensation coefficient is used to adjust the effect of the lag effect on the spray flow rate. This is the hysteresis attenuation coefficient, representing the intensity of the device's response delay; To compensate for time, this reflects the time required from the current change in nozzle opening to the stabilization of the spray flow rate.

[0043] In this implementation plan, the compensation time is calculated by acquiring the spray compensation value, the nozzle opening adjustment time, and the spray flow stabilization time. An exponential decay is constructed by introducing a hysteresis response attenuation coefficient to generate a hysteresis response correction term. Combined with the hysteresis compensation coefficient dynamically updated by the feedback control and optimization module and stored in the operation control database, the hysteresis compensation value is finally calculated. This achieves quantitative assessment and dynamic compensation of the response delay of the spraying equipment, effectively corrects the time deviation between nozzle opening adjustment and actual flow output, ensures that the spraying action and flight control command remain synchronized, and provides a quantitative basis for identifying abnormal response delay sections and triggering advanced control strategies.

[0044] Specifically, the steps for performing spray output timing correction and abnormal section control operations based on the lag analysis results are as follows: By comparing the lag compensation value and the lag threshold in real time, when the lag compensation value is less than the lag threshold, the spray response is stable, the output and control commands remain synchronized, and the current spraying state continues to operate; when the lag compensation value is greater than or equal to the lag threshold, a first-order lead control algorithm with a time constant of τ and a gain of l is used to correct the spray output timing, shorten the response delay of the nozzle and pump, and eliminate the time deviation between the spraying action and flight control. If the lag compensation value calculated after n consecutive rounds of treatment is still greater than or equal to the lag threshold, where n is a preset positive integer, ranging from 3 to 5, with a default value of 3, and each round corresponds to one iteration of the lead control algorithm, then the current operating section is marked as an abnormal response delay section and transferred to the deviation sample queue. At the same time, the lag compensation coefficient is frozen for adaptive updating, keeping the current lag compensation coefficient value in the operation control database unchanged. Figure 4 The figure shows a comparison between the exponential lag compensation and the pure lag step response in this embodiment. By comparing the two step response curves, the control effect of the lag compensation module of this invention is presented intuitively. This verifies that the exponential lag compensation formula can effectively quantify the response delay characteristics of the equipment, and provides a reliable basis for the precise intervention of the advanced control algorithm and the intelligent identification of abnormal sections.

[0045] In this implementation plan, dynamic monitoring and intelligent handling of spraying response delay are achieved by comparing the lag compensation value and the lag threshold in real time: when the delay is below the lag threshold, stable operation is maintained; when the threshold is reached or exceeded, a first-order advance control algorithm is used to actively correct the spraying sequence and eliminate the response delay; if the lag threshold is still exceeded after multiple rounds of compensation, the current section is marked as having an abnormal response delay and transferred to the deviation sample queue, while the lag compensation coefficient update is frozen to ensure parameter stability under abnormal conditions, thereby achieving closed-loop intelligent control of the response delay.

[0046] Specifically, the steps for integrating multi-module operational data and historical deviation samples to conduct intelligent optimization learning and iteratively update control parameters are as follows: Flight trajectory coordinates, wind speed, and spray flow rate data are used as basic inputs. Based on flight control records and environmental sensor records, a cumulative distribution deviation record, deviation area grid markers, and deviation dominant factor ranking records are generated in the operation control database. These are combined with path optimization values, spray compensation values, lag compensation values, and historical deviation samples to form a deviation analysis input set. A Bayesian optimization algorithm is used for learning and training. This algorithm uses a Gaussian process as a surrogate model to construct a prior distribution. Through the balancing exploration and utilization of the acquisition function, the posterior distribution is iteratively updated to approximate the optimal solution of the objective function. The algorithm fits the nonlinear mapping relationship between the cumulative distribution deviation within the deviation area and flight parameters, environmental parameters, and control parameters, ensuring that the predicted deviation remains consistent with the actual observation under various effective operational conditions. The Bayesian optimization algorithm aims to minimize the proportion of the deviation hotspot area to the total area of ​​the operational area and the spray coverage... The variance and the total spray drift outside the boundary are used as the comprehensive optimization targets. The decision variables are a set of compensation coefficients, including speed compensation coefficient, angle compensation coefficient, wind speed compensation coefficient, and lag compensation coefficient, and a path planning weight, including path length weight, wind speed interference weight, flight altitude weight, and spray coverage weight. The initial values ​​of the compensation coefficient set and path planning weight are read from the operation control database and dynamically adjusted within a preset range during the optimization process. At the same time, the pump pressure, spray flow rate, flight speed, and flight altitude are limited to the rated safe operating range specified in the equipment manual, and the area outside the plot boundary is designated as a spraying prohibition zone. Finally, the collaborative optimization of the compensation coefficients and path weights is driven by the distribution characteristics of the deviation hot zone, with the goal of improving the overall spray uniformity. The optimized compensation coefficient set and path planning weight update values ​​are output to realize the synchronous update of the path cost weight, spray compensation coefficient, and lag compensation coefficient. The optimized parameters are written into the operation control database to realize the closed-loop iterative update of the control parameters of each module.

[0047] In this implementation plan, a deviation analysis input set is constructed by integrating flight trajectory coordinates, wind speed, spray flow rate, volume distribution deviation records, deviation area grid markings, deviation dominant factor ranking records, path optimization values, spray compensation values, lag compensation values, and historical deviation samples. A Bayesian optimization algorithm is used to fit the nonlinear mapping relationship between volume distribution deviation within the deviation area and flight parameters, environmental parameters, and control parameters. The comprehensive optimization objective is to minimize the deviation hotspot area, spray coverage variance, and spray drift outside the boundary. The optimization is achieved by collaboratively optimizing the compensation coefficient set (including speed compensation coefficient, angle compensation coefficient, wind speed compensation coefficient, and lag compensation coefficient) and the path planning weight (including path length weight, wind speed interference weight, flight altitude weight, and spray coverage weight). The optimized compensation coefficient set and updated path planning weight values ​​are then written into the operation control database, enabling closed-loop iterative updates of the control parameters for each module and continuously improving spray uniformity and system adaptability.

[0048] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0049] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent spraying control system for agricultural drones, characterized in that, include: The data acquisition and preprocessing module is used to collect multi-source data of UAV operations, preprocess the multi-source data of UAV operations, store it, and build an operation control database. The path planning and optimization module is used to perform a comprehensive cost assessment of the path using trajectory coordinates, environmental sensing, and operational condition data, and to perform adaptive optimization and smoothing control of the flight trajectory based on the cost assessment results. The boundary area dynamic compensation module is used to conduct spraying disturbance analysis based on flight status, wind speed disturbance and terrain feature data, and to perform spraying parameter compensation operations in the boundary area based on the disturbance analysis results; The response lag compensation module is used to conduct execution lag analysis based on timing characteristics, flow status and actuator data, and to perform spray output timing correction and abnormal section control operations based on the lag analysis results. The feedback control and optimization module is used to integrate the operating data of multiple modules and historical deviation samples to carry out intelligent optimization learning and iteratively update the control parameters.

2. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for collecting multi-source data from UAV operations are as follows: Collect multi-source data for drone operations: The onboard positioning module acquires flight trajectory coordinates in real time and calculates the difference in the horizontal coordinates of the trajectory to obtain the horizontal distance of the activity path; the environmental sensors collect wind speed data in real time; the flight control system acquires flight altitude, flight speed changes, nozzle opening adjustment time, path planning start time, and path planning end time; and the image sensor collects farmland prescription maps and spray coverage in real time. The flight angle change is acquired in real time through the inertial measurement unit; the spray flow rate stability is monitored in real time through the flow monitoring sensor, and the moment when the spray flow rate reaches the stability threshold is recorded synchronously; the standard spray flow rate is obtained through the equipment calibration parameters, and the nozzle opening and pump pressure status data of the actuator are recorded in real time.

3. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for preprocessing and storing multi-source data from UAV operations and constructing an operation control database are as follows: Each sensor record is synchronously written with a UAV identifier, acquisition time, task batch number, and geographic coordinate identifier. Time alignment and window aggregation of multi-source data streams are performed based on the acquisition time. Sliding window denoising, outlier removal, and missing value interpolation are applied to the multi-source data of UAV operations. Multi-source records within the same observation time window are integrated into operation observation segments, and each segment is assigned a globally unique segment sequence number. A distribution standardization algorithm is used to uniformly encode and standardize the UAV identifier, task batch number, and operation area identifier, achieving a unified format for the identifier data. A minimum-maximum normalization algorithm is used to normalize and scale the multi-source data of UAV operations to a uniform numerical range. The standardized and normalized multi-source data and observation segments of UAV operations will be supplemented with corresponding segment sequence numbers, UAV identifiers and acquisition time window markers, stored and used to build an operation control database.

4. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for evaluating the comprehensive path cost using trajectory coordinates, environmental sensing data, and operational condition data are as follows: Obtain the horizontal distance of the path at the current location and perform an absolute value operation to obtain the absolute value of the horizontal distance; obtain the wind speed at the current location and perform an absolute value operation to obtain the absolute value of the wind speed; obtain the flight altitude at the current location and perform an absolute value operation to obtain the absolute value of the flight altitude; obtain the original value of the spray coverage and perform an absolute value operation to obtain the absolute value of the spray coverage; sum the above four absolute values ​​to obtain the instantaneous comprehensive cost value; perform a definite integral operation on the instantaneous comprehensive cost value from the start time of path planning to the end time of path planning to obtain the path optimization value.

5. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for performing adaptive optimization and smoothing control of the flight trajectory based on the cost assessment results are as follows: By comparing the path optimization value with the optimization threshold in real time, when the path optimization value is less than the optimization threshold, the existing flight strategy is maintained and operations are carried out continuously and stably. When the optimized path value is greater than or equal to the optimization threshold, first-order partial derivatives are calculated for path length, wind speed interference, flight altitude, and spray coverage, and then sorted in descending order to obtain the path optimization impact queue. Based on the path optimization impact queue, the dominant influencing factor is identified, and corresponding remedial measures are implemented: when the dominant factor is path length, A... The algorithm optimizes path planning and combines cubic B-spline curves to smooth the flight trajectory, reducing turning maneuvers and reciprocating flight. When wind speed is the dominant factor, a dynamic wind field model is constructed based on real-time wind field data collected by the UAV's onboard environmental sensors. The flight trajectory is adjusted in conjunction with the wind field model to reduce the impact of wind on trajectory stability. When flight altitude is the dominant factor, the trajectory elevation change is optimized based on terrain undulations, and the rise and fall amplitude is adjusted smoothly to eliminate the impact of abrupt altitude changes on trajectory smoothness. When spray coverage is the dominant factor, a grid partitioning algorithm is used to divide the work area into m×m grids and calculate the trajectory coverage frequency of each grid. For areas where the coverage frequency is less than the frequency threshold, a greedy algorithm is used to supplement the trajectory path, and an iteration is performed every k seconds until the calculated path optimization value falls back to within the optimization threshold, thereby achieving dynamic optimization and stable control of the flight trajectory.

6. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for conducting spraying disturbance analysis using flight status, wind speed disturbance, and terrain feature data are as follows: Obtain the standard spray flow rate, flight speed variation, flight angle variation, and wind speed; calculate the product of flight speed variation and flight speed compensation coefficient to obtain the speed compensation term; calculate the product of flight angle variation and flight angle compensation coefficient to obtain the angle compensation term; calculate the product of wind speed and wind speed compensation coefficient to obtain the wind speed compensation term; calculate the sum of the speed compensation term, angle compensation term, and wind speed compensation term, and add one to obtain the comprehensive compensation term; multiply the standard spray flow rate value by the comprehensive compensation term to obtain the spray compensation value.

7. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for implementing spray parameter compensation in the boundary area based on the disturbance analysis results are as follows: By comparing the spray compensation value with the spray threshold in real time, when the spray compensation value is less than the spray threshold, the flight attitude is stable, and the existing spray output and travel rhythm are maintained. When the spraying compensation value is greater than or equal to the spraying threshold, the plot boundary is obtained by combining UAV-borne GNSS positioning with farmland prescription maps, and a boundary buffer zone is generated with a distance of B meters from the boundary as the buffer width. The main sources of disturbance are identified by sorting the speed compensation item, the ratio of the speed compensation item to the wind speed compensation item and the comprehensive compensation item in descending order. For disturbances caused by changes in flight speed, a PID control algorithm is used to adjust the flight speed and reduce the speed fluctuation range. For disturbances caused by changes in flight angle, a fuzzy control algorithm is used to optimize the attitude adjustment process, reduce the attitude change amplitude and frequency, and maintain a stable spraying output process. For disturbances caused by wind speed, the wind speed data collected in real time by the airborne environmental sensor is input, and the correlation between wind speed and spraying output is fitted by the least squares method. The output variable is the spraying flow rate adjustment amount, and the spraying output scale is adjusted simultaneously to offset the spraying offset caused by the external wind field until the calculated spraying compensation value falls back to within the spraying threshold. The flight control record is then archived in the operation control database.

8. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for conducting execution lag analysis based on time-series characteristics, flow status, and actuator data are as follows: Obtain the spray compensation value, the time when the spray flow stabilizes, and the time when the nozzle opening is adjusted; calculate the difference between the time when the spray flow stabilizes and the time when the nozzle opening is adjusted to obtain the compensation time; multiply the compensation time by the hysteresis response attenuation coefficient to obtain the time attenuation product term; take the negative value of the time attenuation product term to obtain the negative time attenuation term; perform natural exponentiation on the negative time attenuation term to obtain the exponential attenuation term. Calculate the hysteresis response correction term by subtracting the exponential decay term; obtain the hysteresis compensation coefficient, which is generated by the feedback control and optimization module based on historical deviation samples through iterative optimization and stored in the operation control database; calculate the product of the hysteresis compensation coefficient, the spraying compensation value, and the hysteresis response correction term to obtain the hysteresis compensation value.

9. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for performing spray output timing correction and abnormal section control operations based on the hysteresis analysis results are as follows: By comparing the hysteresis compensation value with the hysteresis threshold in real time, when the hysteresis compensation value is less than the hysteresis threshold, the spraying response is stable, the output and control commands remain synchronized, and the current spraying state is maintained for continuous operation. When the lag compensation value is greater than or equal to the lag threshold, a first-order advance control algorithm with a time constant of τ and a gain of l is used to correct the spray output timing, shorten the response delay of the nozzle and pump, and eliminate the time deviation between the spraying action and the flight control. If the lag compensation value calculated after n consecutive rounds of treatment is still greater than or equal to the lag threshold, the current operation section is marked as an abnormal response delay section and transferred to the deviation sample queue. At the same time, the lag compensation coefficient is frozen for adaptive updating, and the current lag compensation coefficient value in the operation control database remains unchanged.

10. The intelligent spraying control system for agricultural drones according to claim 1, characterized in that: The specific steps for integrating multi-module operational data and historical deviation samples to conduct intelligent optimization learning and iteratively update control parameters are as follows: Flight trajectory coordinates, wind speed, and spray flow rate data are used as basic inputs. Based on flight control records and environmental sensor records, the system generates accumulation distribution deviation records, deviation area grid markers, and deviation dominant factor ranking records in the operation control database. These are combined with path optimization values, spray compensation values, hysteresis compensation values, and historical deviation samples to form the deviation analysis input set. A Bayesian optimization algorithm is used for learning and training to fit the nonlinear mapping relationship between the accumulation distribution deviation within the deviation area and flight, environmental, and control parameters. This ensures that the predicted deviation under various effective operation conditions is consistent with the actual observation. The algorithm minimizes the deviation hotspot area, spray coverage variance, and boundary... The external spray drift is the comprehensive optimization target. The decision variables are a set of compensation coefficients including speed, angle, wind speed, and hysteresis compensation coefficient, and a path planning weight including path length, wind speed interference, flight altitude, and spray coverage weight. At the same time, the pump pressure, flow rate, flight speed, and altitude are limited to the rated safety range of the equipment, and the area outside the boundary is a no-spray zone. The deviation hot zone drives the optimization of compensation coefficients and path weights to improve spray uniformity. The optimized compensation coefficient set is output to update the path cost weight, spray compensation coefficient, and hysteresis compensation coefficient. The optimized parameters are written into the operation control database to realize the closed-loop iterative update of the control parameters of each module.

Images