Unmanned aerial vehicle accurate variable rate fertilization method and system

Abstract

The invention relates to the technical field of agricultural precise management, and discloses an unmanned aerial vehicle precise variable rate fertilization method, which comprises the following steps: collecting vegetation indexes and color feature data of crops in a target plot through an unmanned aerial vehicle carrying a multispectral sensor and an RGB camera; acquiring soil characteristic data of the target land parcel, wherein the soil characteristic data comprises nitrogen, phosphorus and potassium nutrient content, humidity and pH value in soil; the invention further provides an unmanned aerial vehicle precise variable rate fertilization system which comprises a data acquisition module, and the data acquisition module is configured to acquire vegetation indexes, color characteristics, soil characteristics and meteorological data of the target land parcel through a multispectral sensor, an RGB camera, a soil sensor and a meteorological sensor carried by the unmanned aerial vehicle. Through multi-source data fusion, dynamic fertilization adjustment and path optimization, precise variable fertilization is realized, the fertilizer utilization efficiency is improved, the scientificity and efficiency of fertilization operation are improved, and meanwhile, the environmental influence is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural precision management, and specifically to a method and system for precise variable fertilization of unmanned aerial vehicles (UAVs). Background Art

[0002] With the development of modern agricultural production towards intensification and intelligence, the importance of fertilization technology has become increasingly prominent. In traditional fertilization methods, due to the neglect of the spatial differences in crop health status and soil characteristics within the plot, fertilization often shows "averaging" or empirical management, resulting in insufficient fertilization accuracy. In recent years, the precise variable fertilization method based on UAVs has gradually attracted attention. Its core concept is to use the sensors carried by UAVs to comprehensively monitor the plot and combine with the variable fertilization model to dynamically adjust the fertilization strategy to more scientifically and efficiently meet the crop needs.

[0003] In the existing fertilization technologies, the fertilization accuracy and efficiency have been improved through various means. For example, some technical solutions monitor the crop growth status by using multispectral sensors to evaluate the overall health status of the crops within the plot, so as to achieve targeted fertilization. These methods can supplement fertilization for high-demand areas by detecting the soil nutrient content at fixed points and combining with variable fertilization devices, avoiding unnecessary consumption of fertilizer resources. In addition, some technologies use the flexible flight ability of UAVs and combine with path optimization algorithms to improve the efficiency of large-scale farmland fertilization operations, reducing labor costs and energy input.

[0004] However, there are still some deficiencies in the existing technologies. First, most methods are only based on a single data source (such as crop health status or soil nutrient distribution), failing to integrate the advantages of multi-dimensional and multi-source data, resulting in the lack of comprehensiveness and precision of the fertilization strategy. Second, there is a lack of real-time response ability to dynamic environmental conditions such as wind speed and precipitation during the fertilization process, resulting in a greater impact of external factors on fertilization efficiency. Third, the existing path planning schemes fail to fully combine the distribution characteristics of fertilization requirements within the plot, and problems such as insufficient path coverage or repeated fertilization still exist. Finally, the evaluation of the fertilization effect in the existing technologies mainly relies on manual observation or qualitative analysis, lacking scientific and quantitative evaluation means, and at the same time failing to provide clear data support for the optimization of subsequent fertilization strategies. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technologies, the present invention provides a method and system for precise variable fertilization of UAVs, which solves the problems of insufficient fertilization accuracy, lack of real-time dynamic adjustment ability, low efficiency of fertilization path planning, and lack of scientific evaluation of fertilization effect existing in the existing technologies.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for precise variable fertilization of UAVs includes the following steps: Collect vegetation index and color feature data of crops in the target plot through a drone equipped with a multispectral sensor and an RGB camera; Obtain soil property data of the target plot, including the contents of nitrogen, phosphorus, and potassium nutrients, humidity, and pH value in the soil; Collect meteorological data of the target plot, including temperature, humidity, wind speed, and precipitation; Based on the vegetation index data, color feature data, soil property data, and meteorological data of the crops, construct a variable fertilization demand distribution map of the target plot; Estimate the fertilizer requirement at each location in the target plot according to the variable fertilization demand distribution map; Perform variable control on the fertilization device of the drone based on the fertilizer requirement, and implement precise fertilization operations according to the planned path; Collect crop health data in real time and adjust the fertilization parameters of the drone fertilization device; Output a fertilization operation report, recording the fertilizer amount distribution and operation area information.

[0007] Preferably, the collecting of the vegetation index and color feature data of the crops in the target plot includes: Collect the near-infrared reflectance and red light reflectance of the crops in the target plot through a multispectral sensor; Collect the leaf color features of the crops in the target plot through an RGB camera, including green component, red component, and blue component; Calculate the vegetation index and green enhancement index to evaluate the crop health status.

[0008] Preferably, the obtaining of the soil property data of the target plot includes: Collect real-time soil data of the target plot through a soil sensor, including nitrogen, phosphorus, potassium concentrations, humidity, and pH value; Fuse the real-time collected soil data with the historical soil database of the target plot; Estimate the soil data of the unsampled area based on the interpolation algorithm to generate a soil property distribution map of the entire plot.

[0009] Preferably, the collecting of the meteorological data of the target plot includes: Real-time monitor the temperature, humidity, wind speed, and precipitation of the target plot through a meteorological sensor; Obtain weather forecast data and predict the meteorological conditions of the target plot in the short term in the future; Combine the real-time meteorological data with the predicted meteorological data to generate the environmental parameters required for the variable fertilization model.

[0010] Preferably, the constructing of the variable fertilization demand distribution map of the target plot includes: Fuse the vegetation index data, color feature data, soil property data, and meteorological data of the crops; Calculate the fertilizer requirement distribution at each location of the target plot according to the variable fertilization model; Use the interpolation algorithm to expand the sampling point data and generate the variable fertilization requirement distribution map of the target plot.

[0011] Preferably, the estimating the fertilizer requirement at each location of the target plot includes: Combining the target yield and the crop's nutrient requirement to estimate the fertilizer requirement at each location of the target plot; Modify the variable fertilization model and dynamically adjust the fertilizer requirement based on the crop health data; Determine the high-priority locations in the fertilization requirement area and optimize the fertilization order.

[0012] Preferably, the variable control of the fertilization device of the drone based on the fertilizer requirement and the implementation of the precise fertilization operation according to the planned path include: Plan the flight path of the drone according to the variable fertilization requirement distribution map; Dynamically control the spraying rate and spraying range of the fertilization device; Adjust the flight altitude of the drone in the complex terrain area to ensure the uniformity of fertilization.

[0013] Preferably, the real-time collection of crop health data and the adjustment of the fertilization parameters of the drone fertilization device include: Real-time collect crop health data through the sensors carried by the drone, including the vegetation index and the change of leaf color; Update the fertilization model based on the real-time feedback data and dynamically adjust the parameters of the fertilization device; Avoid missing the areas of repeated spraying or insufficient fertilization.

[0014] The present invention also provides a drone precise variable fertilization system, including: A data collection module configured to collect the vegetation index, color features, soil properties, and meteorological data of the target plot through the multispectral sensor, RGB camera, soil sensor, and meteorological sensor carried by the drone; A data processing module configured to construct the variable fertilization requirement distribution map of the target plot based on the collected data; A flight control module configured to plan the flight path of the drone according to the variable fertilization requirement distribution map; A variable fertilization module configured to estimate the fertilizer requirement based on the requirement distribution map and dynamically adjust the spraying parameters of the fertilization device; A real-time feedback module configured to real-time collect crop health data and adjust the fertilization parameters; A data output module configured to generate a fertilization operation report.

[0015] Preferably, the data processing module includes: Construct a variable fertilization model based on the crop health data and soil property data of the target plot; Generate a fertilization requirement distribution map of the target plot according to the variable fertilization model; Call the flight control module to generate a flight path instruction for the drone.

[0016] The present invention provides a method and system for precise variable fertilization of drones. It has the following beneficial effects: 1. By integrating crop health data, soil property data, and meteorological parameters, the present invention constructs a variable fertilization requirement distribution map with multi-source data fusion. By scientifically estimating and dynamically adjusting the fertilization amount, precise and uniform fertilization is achieved. Different from traditional fertilization techniques that rely on single parameters or empirical estimates, the present invention effectively solves the problems of uneven fertilization and low efficiency, and is especially suitable for precise management of complex plots and different crop types.

[0017] 2. By combining real-time feedback with an environmental correction formula and adjusting fertilization parameters according to crop requirements and weather changes, the present invention significantly improves the fertilizer utilization efficiency. Compared with existing fixed fertilization schemes, it solves the problems of fertilizer waste and environmental pollution caused by excessive fertilization. While reducing the fertilizer usage amount, the crop yield and health status are simultaneously improved.

[0018] 3. Through the fertilization amount distribution map, fertilizer utilization efficiency analysis, and crop health improvement map, the present invention realizes the visualization and scientific evaluation of the entire fertilization operation process. This method is convenient for intuitively judging the fertilization effect and provides detailed data support for optimizing subsequent fertilization strategies. Compared with traditional technical solutions lacking quantitative analysis and traceability capabilities, it greatly enhances the scientificity and reliability of fertilization management. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flowchart of the method of the present invention; Figure 2 is a system structure diagram of the present invention; Figure 3 is a system structure diagram of the data acquisition module of the present invention; Figure 4 is a system structure diagram of the data processing module of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0020] Next, in combination with the accompanying drawings in the specification of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to the attached Figure 1 , the embodiments of the present invention provide a method for precise variable fertilization of unmanned aerial vehicles, including the following steps: S1. Collection of crop health data: Using an unmanned aerial vehicle equipped with a multispectral sensor and an RGB camera to collect vegetation index and color feature data of crops in the target plot; The main purpose of this step is to obtain key indicators reflecting the health status of crops, including vegetation index and color feature data. Generally, healthy crops show high consistency in spectral reflection characteristics and color distribution. Therefore, by combining multispectral and RGB data, the growth state of crops can be comprehensively characterized.

[0022] In this embodiment, the process of collecting crop health data includes the following contents: Collection and processing of multispectral data Generally, a multispectral sensor can capture reflectance data in the near-infrared and red light bands of crops, and these bands are particularly critical for evaluating the health status of crops. Specifically, based on these band data, the normalized difference vegetation index can be calculated, which is used to quantify the health status of crops in the target plot. The calculation formula is as follows:

[0023] Where: : Normalized difference vegetation index, the range is usually from -1 to 1, and the value of healthy vegetation is usually close to 1, while the value of non-vegetation areas (such as soil or withered vegetation) is close to 0 or lower; : Reflectance in the near-infrared band of the target plot, reflecting the photosynthesis intensity of vegetation. Healthy vegetation has a high reflectance to near-infrared light; : Reflectance in the red light band of the target plot. Healthy vegetation has a strong absorption of red light and thus a lower reflectance.

[0024] In a possible implementation, the multispectral sensor is installed below the unmanned aerial vehicle to ensure vertical scanning of the plot. The unmanned aerial vehicle sequentially obtains the near-infrared and red light reflection data of different areas of the crop according to the planned path, and generates a pixel-level distribution map in combination with the plot distribution.

[0025] RGB Image Acquisition and Color Enhancement: As a supplement to multispectral data, RGB cameras play an important role in data acquisition. Specifically, RGB cameras can capture color feature data of crop leaves, including the green component ( ), the red component ( ), and the blue component ( ). Through these color components, the green enhancement index ( ) can be further calculated to strengthen the assessment of crop health status. The formula is as follows:

[0026] Where: : Green enhancement index, reflecting the green degree of the target crop. A higher value usually indicates a healthier crop; : Pixel value of the green component in the RGB image. The health of green leaves is usually determined by the reflection intensity of green; : Pixel value of the red component in the RGB image; : Pixel value of the blue component in the RGB image.

[0027] Generally, healthy leaves have a higher green enhancement index, while yellowish or diseased leaves have a lower green enhancement index. In a possible implementation, the RGB camera works in cooperation with the multispectral sensor to form multi-dimensional data acquisition of the plot crops. In addition, the data collected by the RGB camera can be further used for texture analysis, such as leaf lesion detection or pest assessment.

[0028] As an option, after the data and data collected in this embodiment are initially calculated, they will be comprehensively processed through a data fusion algorithm. Specifically, the multispectral data and RGB data are fused in a weighted manner to form a more comprehensive characterization of crop health status. The fusion formula is as follows:

[0029] Where: : Comprehensive crop health index at position ; : Normalized difference vegetation index at position ; : Green enhancement index at position ; and : Weight coefficients, reflecting the relative importance of and in health assessment. The weight values can be adjusted according to the actual crop variety and health status.

[0030] In some embodiments, the weight coefficients are dynamically determined by a training model. For example, for crops that are more sensitive to changes in leaf color, the value can be increased so that the green enhancement index has a higher weight in health assessment; In a possible application scenario, a drone covers a target plot of 5 hectares along a preset flight path. The multispectral sensor acquires the near-infrared and red light reflectance once per second, and the RGB camera captures leaf color images at an interval of 0.5 seconds. After completing one flight, the system generates the distribution map, distribution map, and the comprehensive health index distribution map of this plot. These data will serve as the basis for constructing the variable fertilization requirement distribution map and provide a direct basis for the subsequent fertilization model.

[0031] Time and space resolution processing of data Generally, and the resolution of RGB data can be determined by the flight altitude of the drone, the sensor sampling frequency, and the image pixel size. As an implementation, when the flight altitude of the drone is 20 meters, the and RGB image resolution is 0.5 square meters per pixel. Through the block processing technology, the health data is divided into unit data blocks in the form of a grid, and each block corresponds to a fertilization target point.

[0032] In some embodiments, to improve the temporal consistency of the data, and data will be interpolated in time series after acquisition to eliminate fluctuations caused by environmental factors such as changes in light and wind speed; S2. Acquisition of soil property data: Acquire the soil property data of the target plot, including the contents of nitrogen, phosphorus, and potassium nutrients, humidity, and acidity in the soil; Generally, the fertilizer requirement of crops is closely related to the existing nutrient content in the soil. Therefore, accurately grasping the soil properties (such as the contents of nitrogen, phosphorus, and potassium nutrients, humidity, and acidity, etc.) in the plot is of great significance for the subsequent calculation of fertilization requirements and the optimization of the model. The core of this step is to comprehensively collect and analyze the soil data of the target plot through the soil sensor carried by the drone to provide the necessary basic data for constructing the variable fertilization requirement distribution map.

[0033] In this embodiment, the acquisition and processing of soil property data specifically include the following content: Generally, the drone is equipped with a soil sensor and can collect the soil property data of the target plot in real time during low-altitude flight. Specifically, the soil sensor can detect the following key parameters: The nitrogen, phosphorus, and potassium (N, P, and K) nutrient concentrations represent the contents of nitrogen, phosphorus, and potassium in the soil, usually in units of mg / kg; Soil moisture reflects the water content in the soil, usually in percentage; Soil acidity and alkalinity (pH value) is used to evaluate the acidity and alkalinity of the soil, and the suitable range is usually from 5.5 to 7.5.

[0034] In a possible implementation, the drone conducts point-by-point collection along a preset path. At fixed intervals, the soil sensor emits a weak current to the ground and measures the soil property data of the target area based on the returned signal. The distribution density of the collection points can be adjusted according to the plot area and crop type. For example, for crops with a higher planting density, the interval between sampling points can be appropriately reduced to improve the spatial resolution of the soil data.

[0035] Generally, the soil data of a plot is not evenly distributed, and there may be unsampled points in some areas. Therefore, the collected soil data needs to be processed by an interpolation algorithm to estimate the soil properties of the unsampled areas.

[0036] Specifically, the Kriging interpolation algorithm is used to spatially expand the soil data, and its basic formula is as follows:

[0037] Where: : The soil property value (such as nitrogen concentration or moisture) at the target location; : The soil property value at the sampling point : The sampling point : The soil property value; : The interpolation weight, indicating the contribution of the sampling point to the target location : Usually calculated based on the spatial distance between the sampling point and the target location; : The number of sampling points participating in the interpolation calculation.

[0038] In a possible implementation, the calculation of the weight coefficient follows the following formula:

[0039] Where: : The spatial distance between the target location and the sampling point, usually in meters; : The square of the spatial distance, indicating that sampling points farther away have less influence on the target location.

[0040] Through the above interpolation algorithm, a soil property distribution map of the target plot can be generated. As an option, this distribution map is stored in a grid form, where each grid corresponds to a spatial unit (such as per square meter) and contains complete soil property information.

[0041] In some embodiments, to further improve the accuracy and integrity of soil property data, the soil data collected in real time in this embodiment is fused with the historical database of the target plot. The historical database usually includes information such as soil monitoring data, crop planting records, and fertilization conditions of the plot over the years.

[0042] Specifically, the fusion process can be achieved by the weighted average method, and its formula is as follows:

[0043] Where: : The comprehensive soil property value at the target location; of; : The soil property data collected in real time; : The soil property data in the historical database; and : The weight coefficients, reflecting the relative importance of real-time data and historical data. Generally, the weight of real-time data is higher; Through the above fusion method, the noise in the real-time data can be further corrected, and the historical property data in the unsampled area can be supplemented, making the soil property distribution map more comprehensive and reliable; In a possible application scenario, when the soil data of the target plot changes significantly due to rainfall, irrigation, or soil erosion, the soil sensors in this embodiment can re-collect key data in real time and correct the original distribution map through a dynamic update algorithm; For example, after continuous rainfall, the soil humidity may increase significantly, while the nutrient concentration may decrease due to runoff effects. In this case, the dynamic update algorithm will adjust the corresponding parameter values in the soil distribution map in real time according to the new collected data to ensure the accuracy of the fertilization model; In a practical application, a drone collects soil data for a plot with an area of 10 hectares. The sensors are arranged at sampling points at intervals of 20 meters. The collected nitrogen concentration data ranges from 15 to 25 mg / kg, the humidity ranges from 25% to 35%, and the pH value ranges from 6.0 to 7.0. Through the Kriging interpolation algorithm, the soil properties in the unsampled area are estimated, and a complete soil property distribution map is generated in combination with the historical database data. These data will be directly used for the subsequent estimation of fertilizer requirements and the optimization of the fertilization model; S3. Meteorological data collection: Collect the meteorological data of the target plot, including temperature, humidity, wind speed, and precipitation; In this step, real-time meteorological data of the target plot is collected by the meteorological sensors carried by the unmanned aerial vehicle (UAV), and combined with the predicted data of the weather forecast platform, providing the necessary environmental parameter support for the construction and dynamic adjustment of the subsequent variable fertilization demand distribution map.

[0044] In this embodiment, the specific content of meteorological data collection and processing includes the following aspects: Generally, the UAV can conduct real-time monitoring of the meteorological conditions of the target plot through the carried meteorological sensors. Specifically, the collected data includes the following: Temperature: The real-time air temperature of the plot, in degrees Celsius (°C), used to evaluate the activity level of crop growth and metabolic activities; Humidity: The relative humidity in the air of the plot, in percentage (%), used to reflect the evaporation intensity and leaf surface water balance; Wind speed: The wind speed above the plot, in meters per second (m / s), and the wind speed directly affects the uniformity of fertilization spraying; Precipitation: The current or recent rainfall of the plot, in millimeters (mm), used to judge whether there is a risk of fertilizer loss.

[0045] In a possible implementation, the meteorological sensors collect data at a cycle of every minute, and the distribution density of data points is synchronized with the UAV flight path. As an option, these meteorological parameters are associated with the geographical coordinates of the plot to generate a spatialized meteorological distribution map.

[0046] In some embodiments, in order to improve the dynamic adaptability of the fertilization model, this embodiment also combines weather forecast data. Specifically, the weather forecast data mainly includes the short-term meteorological change trends within the next 1 to 3 days for the target plot, such as the expected temperature change, rainfall possibility and intensity, etc.

[0047] In a possible implementation, the real-time collected meteorological data and the weather forecast data are integrated together by the method of weighted fusion. The fusion formula is as follows:

[0048] Where: : The target location The comprehensive meteorological parameter value at time ; : The real-time collected meteorological data; : The weather forecast data; : The weighting coefficient, indicating the relative importance of the real-time data and the predicted data, with a range of 0 to 1.

[0049] Generally, when the time range is short (such as several hours), the weight of the real-time data is higher, for example = 0.8. When it is necessary to predict longer-term meteorological changes, the weight of the weather forecast data will be appropriately increased; Meteorological conditions are not only input parameters of the variable fertilization model but may also have a direct corrective effect on fertilization requirements. For example, precipitation and wind speed affect the distribution efficiency of fertilizers in the soil, so it is necessary to introduce a correction factor into the fertilization model. Specifically, the formula for fertilization amount correction is as follows:

[0050] Where: : The target location At time The corrected fertilization amount; : The target location At time The original fertilization amount; : The target location At time The rainfall; : The target location At time The wind speed; : The rainfall correction coefficient, ranging from 0 to 1, representing the reduction ratio of rainfall on fertilization requirements; : The wind speed correction coefficient, ranging from 0 to 1, representing the impact of wind speed on fertilization uniformity.

[0051] Specifically, when the rainfall is large, The value of

[0052] will be significantly reduced to avoid fertilizer loss. Similarly, under high wind speed conditions, the spraying rate and angle of the fertilization device will be adjusted to ensure that the fertilizer can evenly cover the target area.

[0053] To further analyze the impact of meteorological data on fertilization effects, in this embodiment, the collected meteorological data is stored as a time series and combined with the terrain data of the target plot to construct a three-dimensional space model. For example, through the vertical profile analysis of temperature, humidity, and wind speed, the fertilization adaptability of different plot areas can be accurately evaluated.

[0054] In a practical application, a drone collects meteorological data for a plot of land with an area of 15 hectares. The results show that the temperature range is from 22°C to 28°C, the humidity range is from 50% to 65%, the wind speed range is from 2 m / s to 5 m / s, and the rainfall in a local area reaches 10 mm. Combining with the weather forecast data, the system predicts that the wind speed of this plot may increase to 7 m / s and the rainfall may further increase within the next 3 hours. Based on these data, the fertilization model reduces the fertilization amount in the high-rainfall area and adjusts the spraying angle of the drone spraying device; S4. Construct a variable fertilization demand distribution map: Based on the vegetation index data, color feature data, soil property data, and meteorological data of the crop, construct a variable fertilization demand distribution map of the target plot; This step constructs a distribution map that can intuitively express the fertilization demand based on the previously collected data, using a multi-source data fusion algorithm and spatial modeling technology, providing clear guidance for subsequent fertilization operations.

[0055] In this embodiment, the specific content of constructing the variable fertilization demand distribution map includes the following aspects: Generally, the generation of variable fertilization demand needs to comprehensively consider crop health data, soil property data (such as nitrogen, phosphorus, potassium concentrations, and humidity), and meteorological data (such as precipitation and wind speed). There is a certain correlation between these data, so joint analysis needs to be carried out through a fusion algorithm. As a possible implementation, this embodiment uses a weighted linear model to fuse multi-source data, and its formula is as follows:

[0056] Where: : Location The variable fertilization demand value at, reflecting the fertilizer requirement intensity at this location; : The normalized difference vegetation index in the crop health data; : The green enhancement index in the crop health data; : Soil property data, including comprehensive indicators such as nitrogen concentration and humidity; : Meteorological data, including precipitation and wind speed, etc.; , , , : Weight coefficients, respectively representing the relative importance of each data item in the fertilization demand.

[0057] Specifically, the weight coefficients can be dynamically adjusted according to the crop type and plot characteristics. For example, in areas with better crop health conditions, The weight of can be appropriately reduced; while in plots with complex meteorological conditions, the weight of meteorological data needs to be increased; In a possible implementation, the calculation formula for variable fertilization requirements based on multi-source data fusion is as follows:

[0058] Where: : Location The variable fertilization requirement at position, in kg / ha; : Location The variable fertilization requirement value at position (calculated by the aforementioned formula); : Fertilization coefficient, used to adjust the fertilizer requirements of different crops; : Location The existing soil nutrient content at position, including nitrogen, phosphorus, potassium, etc.; : The correction coefficient of the existing soil nutrients, indicating the offset effect of the existing soil nutrients on the fertilization requirement.

[0059] Specifically, The value of can be set according to the target crop type and yield target. For example, for corn that requires a large amount of nitrogen fertilizer, The value can be set to 1.2; while for rice with relatively low requirements, The value can be set to 0.8. Similarly, The value of can be adjusted according to the soil test results to reflect the impact of the existing nutrients on the fertilization requirement.

[0060] Generally, the variable fertilization requirement distribution map is spatially displayed in units of plots, facilitating the subsequent fertilization path planning of drones. In this embodiment, the Kriging interpolation algorithm is used to expand the demand data of discrete sampling points, and its formula is as follows:

[0061] Where: : The target position The variable fertilization requirement at; : The sampling point The variable fertilization requirement at; : The interpolation weight, determined by the spatial distance between the sampling point and the target position; : The number of sampling points participating in the interpolation calculation.

[0062] In a possible implementation, the calculation of the interpolation weight follows the following formula:

[0063] Where: : The spatial distance between the target position and the sampling point , in meters.

[0064] Through the Kriging interpolation algorithm, the variable fertilization demand values at each location within the plot can be generated and stored in the form of a raster map, with each raster cell corresponding to a spatial unit (such as per square meter or per 10 square meters). As an option, the resolution of the distribution map can be adjusted according to the spraying accuracy of the drone fertilization device.

[0065] In some embodiments, to facilitate the manager's intuitive understanding of the variable fertilization demand distribution, this embodiment also outputs the distribution map in an image format. Specifically, the distribution map uses color gradients to represent the fertilization intensity in different regions. For example: Red represents the area with high fertilizer demand; Yellow represents the area with medium fertilizer demand; Green represents the area with low fertilizer demand.

[0066] In addition, the output format of the distribution map can include forms such as two-dimensional plane maps and three-dimensional topographic maps, which are suitable for different analysis requirements.

[0067] In a practical application, a drone conducts a variable fertilization demand analysis on a plot with an area of 20 hectares. Combining crop health data, soil property data, and meteorological data, a variable fertilization demand distribution map is generated. The results show that the fertilization demand is relatively high in the area close to the low-lying area of the plot, while the fertilization demand in the slope area is relatively low. The distribution map is output in units of per square meter with a resolution of 0.5 meters, serving as a direct input for subsequent fertilization path planning; S5. Fertilizer demand estimation: According to the variable fertilization demand distribution map, estimate the fertilizer demand at each location in the target plot; Generally, the calculation of fertilizer demand not only depends on the variable fertilization demand distribution but also needs to comprehensively consider the combined effects of crop growth stage, existing soil nutrients, and environmental conditions. Therefore, the calculation of fertilizer demand provides a direct guiding basis for subsequent precise fertilization operations.

[0068] In this embodiment, the specific content of fertilizer demand estimation includes the following aspects: Generally, the basic estimation of fertilizer demand takes the demand value in the variable fertilization demand distribution map as the core input. As a possible implementation method, this embodiment uses the following formula for basic estimation:

[0069] Where: : Location The basic fertilization demand at, in kg / ha; : The fertilization coefficient of the target crop, used to reflect the nutrient absorption efficiency of the crop, usually determined by experimental data or planting experience; : Location The variable fertilization requirement value is calculated by the formula in step S4.

[0070] As an option, The value of can be dynamically adjusted according to the crop type and growth stage. For example, for crops with high absorption efficiency (such as corn), the value of can be set to 0.9; while for crops with relatively low absorption efficiency (such as wheat), the value of may need to be reduced to 0.7.

[0071] The fertilizer requirement obtained by the basic estimation also needs to be corrected in combination with the existing soil nutrient data to avoid waste or environmental pollution caused by excessive fertilization In some embodiments, in order to ensure that the variable fertilization requirement matches the target yield, this embodiment further optimizes and adjusts the fertilizer requirement in combination with the target yield. The optimization formula for the fertilizer requirement related to the target yield is as follows:

[0072] Where: : Location The optimized fertilization requirement at, in kg / ha; : Location The corrected fertilization requirement at ; : Target yield, in kg / ha; : Current location The current crop yield at, in kg / ha; : Target yield correction coefficient, used to adjust the influence of the yield gap on the fertilizer requirement.

[0073] Specifically, when the target yield is significantly higher than the current yield, the value of will relatively increase to increase the fertilization intensity; while in the area where the yield is close to the target yield, the value of can be appropriately reduced to reduce fertilizer waste.

[0074] To improve the dynamic adaptability of the fertilizer requirement, this embodiment also adjusts the fertilizer requirement in real time in combination with meteorological conditions. For example, in high rainfall areas, the risk of fertilizer loss is high, so the fertilizer requirement needs to be reduced; while in low rainfall or drought conditions, the fertilizer requirement needs to be appropriately increased to ensure crop growth.

[0075] The adjustment formula is as follows:

[0076] Where: : Location The final fertilizer requirement at, in kg / ha; : Optimized fertilization requirement; : Location The rainfall at the location, in mm; : Location The wind speed at the location, in m / s; : Rainfall correction coefficient, indicating the reduction effect of rainfall on fertilization requirements; : Wind speed correction coefficient, indicating the impact of wind speed on fertilization efficiency.

[0077] In a practical application, for a plot with an area of 10 hectares, the initial data for estimating fertilizer requirements includes the variable fertilization demand distribution map, the existing soil nutrient distribution map, and the target yield data. Calculated through the above formula: the fertilizer requirement in the high-nutrient area of the plot is 20 kg / ha, while the fertilizer requirement in the low-nutrient area reaches 40 kg / ha. After considering the rainfall and wind speed conditions, the fertilizer requirement in some low-lying areas is further reduced to 18 kg / ha to avoid fertilizer loss; S6. Fertilization path planning and variable control: Based on the fertilizer requirement, variable control is performed on the fertilization device of the unmanned aerial vehicle (UAV), and precise fertilization operation is carried out according to the planned path; This step takes the estimated fertilizer requirement as the core basis, combines the variable fertilization demand distribution map, formulates the fertilization path planning scheme for the UAV, and adjusts the spraying rate and range of the fertilization device in real time through dynamic variable control technology. Generally, the design of path planning and variable control needs to comprehensively consider the fertilization demand distribution, the topographic characteristics of the plot, and the flight ability of the UAV to ensure the uniformity and economy of fertilization.

[0078] In this embodiment, the specific content of fertilization path planning and variable control includes the following aspects: Generally, fertilization path planning needs to determine the flight path of the UAV so that it can cover all fertilization required areas and minimize the waste of repeated flight and spraying. Specifically, in this embodiment, the shortest path algorithm is used to plan the flight path of the UAV in combination with the variable fertilization demand distribution map. The core formula is as follows:

[0079] Where: : The total length of the flight path, in meters; : Adjacent location points on the UAV flight path; : Path point and The distance between, in meters; : The total number of path points.

[0080] Specifically, the flight path of the drone will preferentially cover areas with high fertilization requirements while avoiding obstacles and areas that do not require fertilization. In one possible implementation, the plot is divided into several grid cells, and each grid cell corresponds to a variable fertilization requirement value. The path planning algorithm generates a path that efficiently covers all the fertilizer-required cells based on the levels of these requirement values.

[0081] As an option, path planning can also be optimized by combining the terrain data of the plot. For example, in areas of the plot with a large slope, the flight height and speed of the drone can be appropriately adjusted to ensure the uniformity and safety of fertilization.

[0082] After the fertilization path is determined, the fertilization device of the drone needs to perform variable control according to the specific fertilization requirements at each location. Generally, variable control includes the dynamic adjustment of the spraying rate and the spraying range. In this embodiment, the basic formula for variable control is as follows:

[0083] Where: : The spraying rate at position , in kg / s; : The final fertilization requirement at position , in kg / ha, calculated by step S5; : The coverage range per unit area of the spraying device, in ha.

[0084] Specifically, when the drone reaches a certain grid cell, its fertilization device will read the fertilization requirement data of this cell in real time and adjust the spraying amount according to the spraying rate formula. For example, for areas with a high fertilization requirement, the spraying rate will be appropriately increased; while in areas with a low fertilization requirement, the spraying rate will be decreased.

[0085] To further improve the fertilization efficiency, in this embodiment, the spraying range of the fertilization device is also dynamically adjusted in combination with the fertilization requirement data. The adjustment formula for the spraying range is as follows:

[0086] Where: : The spraying range at position , in meters; : The default spraying range of the spraying device, in meters; : The final fertilization requirement at position ; : The average fertilization requirement of the target plot; : The spraying range adjustment coefficient, used to adjust the influence degree of the fertilization requirement on the spraying range.

[0087] Specifically, when the fertilizer requirement of a certain area is significantly higher than the average value of the plot, its spraying range will be correspondingly expanded to improve the spraying efficiency; while in the area with lower fertilizer requirement, the spraying range will be reduced to reduce fertilizer waste.

[0088] Influence of environmental factors on variable control In this embodiment, in order to adapt to complex plot environments and meteorological conditions, the variable fertilization control will also be dynamically corrected in combination with real-time data such as wind speed and rainfall. For example, when the wind speed is high, the spraying range needs to be appropriately reduced to avoid the fertilizer being blown away by the wind; when the rainfall is large, the spraying rate needs to be reduced to reduce the risk of fertilizer loss.

[0089] The environmental correction formula is as follows:

[0090] Where: : Spraying rate after environmental correction; : Original spraying rate; : Wind speed, unit: m / s; : Rainfall, unit: mm; : Wind speed correction coefficient; : Rainfall correction coefficient.

[0091] In a possible implementation, the drone will collect real-time wind speed and rainfall data during operation and input them into the above formula to dynamically adjust the spraying rate and range.

[0092] In a practical application, the drone performs fertilization operations on a target plot with an area of 8 hectares. The default spraying range of the spraying device is 5 meters, and the default spraying rate is 2 kg / s. Through the path planning algorithm, the total flight path length of the drone is optimized to 6000 meters, which is 20% less than the traditional path. At the same time, during the fertilization process, the variable control system adjusts the spraying rate and range according to the fertilizer requirements of high-demand areas and low-demand areas. The results show that the spraying rate in high-demand areas is increased to 3 kg / s, and the spraying range is expanded to 6 meters; the spraying rate in low-demand areas is reduced to 1.5 kg / s, and the spraying range is reduced to 4 meters; S7. Real-time feedback and dynamic adjustment: Real-time collect crop health data and adjust the fertilization parameters of the drone fertilization device; In this step, the drone collects real-time crop health data, fertilization execution status, and environmental change parameters during operation, and uses these data to dynamically adjust the parameters of the fertilization device. Generally, the real-time feedback mechanism combines the variable fertilization model and control algorithm, which can effectively respond to local changes in plot conditions and further improve the accuracy of fertilization and the utilization efficiency of fertilizers.

[0093] In this embodiment, the specific content of real-time feedback and dynamic adjustment includes the following aspects: Generally, when the unmanned aerial vehicle (UAV) is performing fertilization operations, it is equipped with a multispectral sensor and an RGB camera to monitor crop health data in real time. Specifically, in this embodiment, the following two core indicators are collected in real time: Normalized Difference Vegetation Index ( ): It is used to quantify the health status of crops. The real-time calculation formula is as follows:

[0094] Where: : The target position At time The real-time value; : The target position At time The near-infrared band reflectance; : The target position At time The red band reflectance.

[0095] Green Enhancement Index: As a supplementary indicator, the real-time calculation formula is as follows:

[0096] Where: : The target position At time The real-time green enhancement index; , and : Respectively represent the real-time pixel values of the green, red, and blue channels.

[0097] In a possible implementation, the UAV obtains the above data in real time for each fertilization unit at a sampling interval of 2 seconds and compares it with a preset health threshold.

[0098] The health data collected in real time is input into the variable rate fertilization model for dynamically adjusting the spraying rate and range. The adjustment process is based on the following core formula:

[0099] Where: : The dynamically adjusted spraying rate at the target position At time , with the unit of kg / s; : The original spraying rate at position , calculated by the formula in step S6; : The target position At time of change amount.

[0100] Specifically, when the value collected in real time increases significantly, it indicates that the crop health condition has been improved, and the spraying rate will be correspondingly reduced; while when the value decreases or does not change significantly, the spraying rate is appropriately increased to meet the growth requirements of the crops.

[0101] During the fertilization process, the drone will also collect environmental data in real time (such as wind speed and rainfall), and correct the spraying parameters according to these data; Specifically, when the wind speed and rainfall are large, the spraying rate will be significantly reduced to reduce fertilizer loss; under the conditions of small wind speed and rainfall, the normal spraying rate is maintained.

[0102] In this embodiment, to ensure the efficiency of dynamic adjustment, after the fertilization operation is completed, the drone will store all the data collected in real time into the central control system. These data include: Crop health data; Dynamically adjusted spraying rate and range; Environmental parameters collected in real time (wind speed, rainfall, etc.).

[0103] As a possible implementation, these data will be used as the basis for model optimization of subsequent fertilization operations.

[0104] In a practical application, the drone performs a fertilization operation on the target plot, collects and data in real time, and dynamically adjusts the spraying rate in combination with the wind speed and rainfall. The results show that in the area where the value is significantly improved, the spraying rate is reduced from 2 kg / s to 1.5 kg / s; while in the area where there is no obvious improvement and the wind speed is low, the spraying rate is increased to 2.5 kg / s. After the operation is completed, all the data is uploaded to the central control system for further analysis and optimization; In the area where there is no obvious improvement and the wind speed is low, the spraying rate is increased to 2.5 kg / s. After the operation is completed, all the data is uploaded to the central control system for further analysis and optimization; S8, Generation of fertilization operation report: Output a fertilization operation report, recording the distribution of fertilization amount and operation area information; This step generates a comprehensive and visual fertilization report based on real-time feedback data, fertilization parameters, environmental conditions and operation results. Generally, this report includes core contents such as the distribution of fertilization amount, fertilizer use efficiency and the improvement of crop health.

[0105] In this embodiment, the specific content of generating the fertilization operation report includes the following aspects: Under normal circumstances, the fertilizer application rate distribution map presents the fertilizer application rate of each area intuitively in units of plots. Specifically, in this embodiment, a fertilizer application rate distribution map is generated using dynamic fertilizer application data, and its calculation formula is as follows:

[0106] Wherein, : The total fertilizer application amount at the target position, in kg; : The final spraying rate at the target position, in kg / s, calculated by the environmental correction formula in step S7; : The target position : The corresponding fertilization area at the target position, in square meters, and the calculation formula is as follows: : The target position : The corresponding fertilization area at the target position, in square meters, and the calculation formula is as follows:

[0107] Among them: : The dynamic spraying width, in meters; : The flight speed of the drone, in m / s; : The spraying time of the drone at the target position, in seconds.

[0108] As an option, the output format of the fertilizer application rate distribution map can be a two-dimensional raster map or a three-dimensional solid map, and the specific form can be adjusted according to the needs of the manager. For example, the two-dimensional map uses color gradients to represent the fertilizer application rates of different areas, with red representing high fertilizer application rates and green representing low fertilizer application rates.

[0109] The fertilizer use efficiency (NUE) is one of the core indicators for evaluating the fertilization effect. In this embodiment, it is calculated by the following formula:

[0110] Wherein, : The fertilizer use efficiency of the target plot, in kg yield gain / kg fertilizer application amount.

[0111] : The crop yield gain, in kg / ha, and the calculation formula is as follows:

[0112] Wherein: : The actual yield of the target plot after fertilization, in kg / ha; : The predicted yield of the target plot before fertilization, in kg / ha; : The total fertilizer application amount of the target plot, in kg, calculated from the sum of the fertilizer application amounts at all positions in the fertilizer application rate distribution map.

[0113] In a possible implementation, the NUE is divided into several regional values to evaluate the differences in fertilization effects in different plot areas. For example, for areas with higher fertility, the NUE value may be close to 1, while for areas with low fertility, the NUE value may be less than 0.5.

[0114] To further evaluate the fertilization effect, this embodiment also generates a crop health improvement map. This map is obtained by analyzing the changes in crop health data before and after fertilization, and its core formula is as follows:

[0115] Where, : The crop health improvement value at the target location; : The crop health index value after fertilization at the target location : The target location , which can take the value or value; : The crop health index value before fertilization at the target location .

[0116] Specifically, the crop health improvement map uses color gradients to represent the degree of improvement. For example, red represents areas with significant health improvement, yellow represents areas with general improvement, and green represents areas with no obvious improvement.

[0117] As an option, this embodiment also evaluates the environmental impact by analyzing the potential loss of fertilizers during the fertilization process. The calculation formula for environmental loss is as follows:

[0118] Where, : The environmental loss at the target location, in kg; : The total fertilization amount at the target location : The target location ; : The rainfall at the target location, in mm; : The target location ; : The wind speed at the target location, in m / s; , : The loss correction coefficients for rainfall and wind speed, respectively.

[0119] By calculating the environmental loss, the potential impact of fertilization processes in different regions on the environment can be visually evaluated, thus providing an optimization direction for subsequent fertilization strategies.

[0120] In a practical application, after a drone completed precise variable fertilization on a plot of 15 hectares, a complete fertilization operation report was generated. The results showed that: The total fertilization amount in the high-demand area is 500 kg, the fertilization amount in the low-demand area is 300 kg, and the total fertilization amount is 800 kg.

[0121] The calculated fertilizer use efficiency (NUE) of the target plot is 0.75, indicating that each kilogram of fertilization amount brings a 0.75-kilogram increase in crop yield.

[0122] The crop health improvement graph shows that the average NDVI value in the low-fertility area has increased by 0.15, and the NDVI in the high-fertility area has increased by 0.05.

[0123] The environmental assessment shows that the environmental loss in some high-rainfall areas reaches 5 kg, accounting for 0.6% of the total fertilization amount.

[0124] The above data is output in the form of charts, including the fertilization amount distribution map, fertilizer use efficiency distribution map, crop health improvement graph, and environmental impact assessment graph, providing reliable support for managers to intuitively understand the fertilization effect.

[0125] A drone precision variable fertilization system described below can be correspondingly referred to in relation to a drone precision variable fertilization method described above.

[0126] Please refer to the appendix Figure 2 - Appendix Figure 4 , the present invention also provides a drone precision variable fertilization system, including: A data acquisition module configured to collect vegetation index, color characteristics, soil properties, and meteorological data of the target plot through a multispectral sensor, RGB camera, soil sensor, and meteorological sensor carried by the drone; A data processing module configured to construct a variable fertilization requirement distribution map of the target plot based on the collected data; A flight control module configured to plan the flight path of the drone according to the variable fertilization requirement distribution map; A variable fertilization module configured to estimate the fertilizer requirement based on the requirement distribution map and dynamically adjust the spraying parameters of the fertilization device; A real-time feedback module configured to collect crop health data in real time and adjust the fertilization parameters; A data output module configured to generate a fertilization operation report.

[0127] The system of this embodiment can be used to execute the above method embodiment, and its principle and technical effects are similar, which will not be elaborated here.

[0128] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for precise variable fertilization by unmanned aerial vehicle, characterized in that: The following steps are involved: The vegetation index and color characteristic data of the target plot crops are collected by drones equipped with multispectral sensors and RGB cameras; Obtain soil property data of the target plot, including nitrogen, phosphorus, potassium nutrient content, moisture and pH in the soil; Collect meteorological data of the target plot, including temperature, humidity, wind speed and precipitation; Constructing a variable fertilizer demand distribution map of the target plot based on the vegetation index data, color feature data, soil property data and meteorological data of the crop; According to the variable fertilizer demand distribution map, estimate the fertilizer requirement at each location of the target plot; The fertilization device of the drone is controlled by variables based on the fertilizer requirement, and precise fertilization operations are carried out according to the planned path; Collect crop health data in real time and adjust fertilization parameters of drone fertilization devices; Output fertilization operation report, record fertilizer amount distribution and operation area information.

2. The method for precise variable fertilization using an unmanned aerial vehicle according to claim 1, characterized in that: The vegetation index and color feature data of the target plot of crops are collected including: The near-infrared reflectance and red light reflectance of crops in the target plot are collected through a multispectral sensor; The leaf color characteristics of the target plot crops are collected through an RGB camera, including green, red and blue components; Calculate vegetation index and green enhancement index to assess crop health.

3. The method for precise variable fertilization by unmanned aerial vehicle according to claim 1, characterized in that: The step of obtaining soil property data of the target plot includes: Collect real-time soil data of the target plot through soil sensors, including nitrogen, phosphorus, potassium concentration, moisture and pH; Fusion of real-time collected soil data with the historical soil database of the target plot; The soil data of the unsampled area are estimated based on the interpolation algorithm to generate a soil property distribution map for the entire plot.

4. The method for precise variable fertilization by unmanned aerial vehicle according to claim 1, characterized in that: The meteorological data collected for the target plot includes: Real-time monitoring of temperature, humidity, wind speed and precipitation of target plots through meteorological sensors; Obtain weather forecast data to predict the meteorological conditions of the target plot in the short term in the future; The real-time meteorological data is combined with the predicted meteorological data to generate the environmental parameters required for the variable fertilization model.

5. The method for precise variable fertilization using an unmanned aerial vehicle according to claim 1, characterized in that: The variable fertilization demand distribution map of the target plot is constructed as follows: Integrate crop vegetation index data, color feature data, soil property data and meteorological data; Calculate the fertilizer requirement distribution at each location of the target plot based on the variable fertilization model; The sampling point data were expanded using an interpolation algorithm to generate a variable fertilizer demand distribution map for the target plot.

6. The method for precise variable fertilization by unmanned aerial vehicle according to claim 1, characterized in that: The estimated fertilizer requirement for each location of the target plot includes: Combine the target yield and crop nutrient requirements to estimate the fertilizer requirement for each location in the target plot; Modified variable fertilization model to dynamically adjust fertilizer requirements based on crop health data; Identify high priority locations that require fertilization and optimize the order in which fertilizers should be applied.

7. The method for precise variable fertilization by unmanned aerial vehicle according to claim 1, characterized in that: The variable control of the fertilization device of the UAV based on the fertilizer requirement and the implementation of the precise fertilization operation according to the planned path include: Plan the flight path of the drone based on the variable fertilizer demand distribution map; Dynamically control the spraying rate and spraying range of the fertilization device; Adjust the drone's flight altitude in areas with complex terrain to ensure uniform fertilization.

8. The method for precise variable fertilization using an unmanned aerial vehicle according to claim 1, characterized in that: The real-time collection of crop health data and adjustment of fertilization parameters of the drone fertilization device include: Sensors are used to collect crop health data in real time, including vegetation index and leaf color changes; Update the fertilization model based on real-time feedback data and dynamically adjust the parameters of the fertilization device; Avoid duplicate spraying or omission of under-fertilized areas.

9. A UAV precision variable fertilization system, characterized in that: A method for precise variable fertilization using an unmanned aerial vehicle according to any one of claims 1 to 8, comprising: A data acquisition module, configured to collect vegetation index, color characteristics, soil properties and meteorological data of a target plot through a multispectral sensor, an RGB camera, a soil sensor and a meteorological sensor carried by the UAV; A data processing module configured to construct a variable fertilization demand distribution map of a target plot based on the collected data; a flight control module configured to plan a flight path of the UAV according to the variable fertilizer demand distribution map; A variable fertilization module configured to estimate fertilizer demand based on a demand distribution map and dynamically adjust spraying parameters of a fertilization device; A real-time feedback module configured to collect crop health data in real time and adjust fertilization parameters; A data output module is configured to generate a fertilization operation report.

10. The UAV precise variable fertilization system according to claim 9, characterized in that: The data processing module comprises: Construct a variable fertilization model based on crop health data and soil property data of the target plot; Generate a fertilizer demand distribution map of the target plot based on the variable fertilization model; Call the flight control module to generate the flight path instructions for the drone.

Images