Plant nitrogen fertilizer recommendation method and system based on unmanned aerial vehicle spectrum diagnosis
By using drone-based spectral diagnostics and closed-loop feedback optimization, and dynamically monitoring plant nitrogen uptake, the problems of large deviations in nitrogen fertilizer recommendations and time-consuming and labor-intensive processes in existing technologies have been solved, achieving precise fertilization and efficient utilization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGXI ZHUANG AUTONOMOUS REGION ACAD OF AGRI SCI
- Filing Date
- 2025-12-09
- Publication Date
- 2026-05-08
AI Technical Summary
Existing crop nitrogen fertilizer recommendation technologies rely on static data and fail to dynamically reflect nitrogen demand, resulting in large recommendation biases, making it impossible to achieve precise fertilization. Furthermore, these technologies are time-consuming and labor-intensive, impacting crop growth and the environment.
By using drone spectral diagnostics to dynamically monitor plant nitrogen absorption, the nitrogen absorption curve is corrected. Combined with light and temperature data, a precise nitrogen fertilizer recommendation plan is generated, and the fertilization plan is optimized through closed-loop feedback.
It enables accurate prediction of plant nitrogen requirements, improves nitrogen fertilizer utilization, reduces waste, and protects the ecological environment.
Smart Images
Figure CN121998285A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of plant nitrogen fertilizer management technology, specifically relating to a plant nitrogen fertilizer recommendation method and system based on UAV spectral diagnosis. Background Technology
[0002] Nitrogen is a key limiting element for crop growth, and its supply level directly determines crop yield and quality. Therefore, how to apply nitrogen fertilizer precisely is a current issue in the field of smart agriculture.
[0003] Existing crop nitrogen fertilizer recommendation technologies have defects and limitations. In terms of diagnostic methods, traditional technologies rely on static historical data or empirical models, estimating based on regional average fertilizer application rates or the theoretical saturation uptake of crops. This model ignores the dynamic changes in crop nitrogen requirements during the actual growth cycle and fails to take into account the combined effects of soil conditions, light, temperature, humidity, and different growth stages of crops in specific plots. This leads to a discrepancy between recommended fertilizer application rates and actual crop needs, resulting in poor applicability and accuracy. In terms of execution efficiency and spatial dimension, existing diagnostic methods are often time-consuming, labor-intensive, and costly, and cannot achieve large-area, high-throughput real-time monitoring. It is difficult to conduct differentiated and refined fertilization operations based on the spatial heterogeneity of crop growth in the field. Uniform application not only wastes nitrogen in areas with sufficient nitrogen but also causes yield reduction in nitrogen-deficient areas due to insufficient nutrients. These technical defects may inhibit or reduce the utilization efficiency of nitrogen fertilizer. Unabsorbed nitrogen fertilizer remains in the soil or enters water bodies with surface runoff, leading to eutrophication and increased soil compaction, posing a threat to the ecological environment. Summary of the Invention
[0004] The purpose of this invention is to provide a method and system for recommending nitrogen fertilizer for plants based on UAV spectral diagnosis, in order to solve the problems in the existing technology where there are certain deviations in assessing the ability of plants to absorb nitrogen fertilizer and the recommended nitrogen fertilizer has poor applicability.
[0005] To achieve the above objectives, the technical solution of the present invention is as follows: a method for recommending nitrogen fertilizer for plants based on UAV spectral diagnosis, comprising the following steps: Determine the deviation between historical nitrogen content data and current nitrogen content data of the target plant; based on the deviation between historical nitrogen content data and current nitrogen content data of the target plant, correct the initial nitrogen absorption curve characterizing the historical nitrogen absorption trend to generate a corrected nitrogen absorption curve; Based on the corrected nitrogen absorption curve and the nitrogen requirement information of the target plant, a recommended total amount of nitrogen fertilizer is generated, and the recommended total amount of nitrogen fertilizer is divided into at least one dose to determine the corresponding addition time point. After applying nitrogen fertilizer based on the addition time point, the effective nitrogen increment corresponding to one unit of nitrogen fertilizer is calculated based on the monitored changes in nitrogen content of the target plant. The addition time point is corrected based on the effective nitrogen increment to generate the corrected addition time point. The steps for generating the recommended total amount of nitrogen fertilizer to be applied based on the corrected nitrogen absorption curve and the nitrogen requirement information of the target plant specifically include: calculating the estimated total amount of nitrogen to be added based on the initial nitrogen absorption upper limit and nitrogen requirement information; calculating the actual requirement of the target plant based on the data in the stable absorption area; when the difference between the estimated total amount of nitrogen to be added and the actual requirement is less than the preset allowable deviation value, determining the recommended total amount of nitrogen fertilizer to be applied based on the estimated total amount of nitrogen to be added; otherwise, determining the recommended total amount of nitrogen fertilizer to be applied based on the actual requirement.
[0006] Preferably, the step of obtaining current nitrogen content data includes: determining an image acquisition path within the drone flight area set for the target plant, acquiring images at multiple image acquisition points by flying the drone along the acquisition path, and processing the images to identify the target plant and obtain current nitrogen content data.
[0007] Preferably, the method further includes: comparing the deviation between historical nitrogen content data and current nitrogen content data with a preset nitrogen screening threshold; if the deviation is greater than or equal to the nitrogen screening threshold, the nitrogen absorption status of the target plant is determined to be abnormal.
[0008] Preferably, the step of correcting the initial nitrogen absorption curve characterizing the historical nitrogen absorption trend to generate a corrected nitrogen absorption curve includes: acquiring light intensity data, effective light-collecting area data, and temperature data of the target plant during its growth cycle; calculating a first nitrogen absorption efficiency based on the light intensity data and the effective light-collecting area data; calculating a second nitrogen absorption efficiency based on the temperature data; determining the average of the first nitrogen absorption efficiency and the second nitrogen absorption efficiency as the overall nitrogen absorption efficiency; and determining the model parameters used to characterize the initial nitrogen absorption curve based on the overall nitrogen absorption efficiency.
[0009] Preferably, the step of generating the recommended total amount of nitrogen fertilizer to be applied based on the modified nitrogen absorption curve and the nitrogen requirement information of the target plant further includes: dividing the growth stage of the target plant into multiple time periods, calculating the nitrogen absorption rate in each time period, and if it is determined that the nitrogen absorption rate continues to decline in at least two consecutive time periods, then the current time period and its subsequent time periods are defined as the stable absorption region.
[0010] Preferably, when the difference between the estimated total amount of nitrogen to be added and the actual demand is greater than or equal to the allowable deviation value, the step of determining the recommended total amount of nitrogen fertilizer to be applied based on the actual demand is replaced by: correcting the estimated total amount of nitrogen to be added based on the difference between the estimated total amount of nitrogen to be added and the actual demand to obtain a preliminary corrected total amount of nitrogen to be added; adjusting the preliminary corrected total amount of nitrogen to be added based on the change between the preliminary corrected total amount of nitrogen to be added and the previously optimized total amount of nitrogen to be added to obtain the current optimized total amount of nitrogen to be added; and determining the recommended total amount of nitrogen fertilizer to be applied based on the current optimized total amount of nitrogen to be added.
[0011] Preferably, the step of determining the addition time point includes: adding an initial offset to the current time to determine the addition time point of the first dose, and accumulating a preset time interval based on the addition time point of the previous dose to determine the addition time point of the next dose.
[0012] A plant nitrogen fertilizer recommendation system based on UAV spectral diagnostics includes the following modules: The nitrogen status monitoring module is used to acquire historical and current nitrogen content data of the target plant. The fertilizer recommendation generation module, in response to the data obtained by the nitrogen status monitoring module, is used to determine the deviation between historical nitrogen content data and current nitrogen content data, and generates a nitrogen fertilizer recommendation scheme based on the deviation, including at least one dosage and the corresponding addition time point. The fertilization effect evaluation module is used to calculate the effective nitrogen increment based on the monitored changes in nitrogen content of the target plants after applying nitrogen fertilizer according to the recommended nitrogen fertilizer plan, and to correct the addition time point based on the effective nitrogen increment to generate the corrected addition time point.
[0013] Preferably, the nitrogen status monitoring module acquires current nitrogen content data by controlling a drone to collect images within the drone flight area of the target plant; processing the images to identify the target plant and acquire its current nitrogen content data.
[0014] Preferably, the fertilization recommendation generation module is also used to: compare the deviation with a preset nitrogen screening threshold; if the deviation is greater than or equal to the nitrogen screening threshold, then the nitrogen absorption status of the target plant is determined to be abnormal. When generating nitrogen fertilizer recommendation plans, the fertilizer recommendation generation module is also used to: calculate the estimated total amount of nitrogen to be added and the actual requirement; and, based on the comparison between the difference between the estimated total amount of nitrogen to be added and the actual requirement and a preset allowable deviation value, selectively determine the recommended total amount of nitrogen fertilizer to be applied based on either the estimated total amount of nitrogen to be added or the actual requirement. Beneficial effects 1. This invention determines the deviation between the historical nitrogen content data and the current nitrogen content data of the target plant, and corrects the initial nitrogen absorption curve based on the deviation to generate a corrected nitrogen absorption curve. This can dynamically calibrate the plant nitrogen absorption model, making it not only dependent on historical trends, but also more accurately reflecting the current actual absorption capacity of the target plant, thereby improving the accuracy of plant nitrogen demand prediction and providing a data basis for subsequent total nitrogen fertilizer calculation.
[0015] 2. When calculating the total amount of nitrogen to be added, this invention compares and verifies the estimated total amount of nitrogen to be added based on the initial nitrogen absorption upper limit with the actual demand calculated based on stable absorption region data. If the difference between the estimated total amount of nitrogen to be added and the actual demand exceeds the allowable deviation value, the actual demand or the value after further iterative correction shall be used to determine the recommended total amount of nitrogen fertilizer to be applied. Through this verification and correction mechanism, the accuracy of the final recommended total amount of nitrogen fertilizer to be applied is ensured, and over- or under-fertilization due to model estimation errors is avoided.
[0016] 3. After applying nitrogen fertilizer, this invention calculates the effective nitrogen increment corresponding to a unit of nitrogen fertilizer based on the monitored changes in nitrogen content of the target plant; and corrects the addition time point based on the effective nitrogen increment to generate a corrected addition time point. Through this feedback correction closed loop, the quantitative evaluation of fertilization effect and adaptive optimization of the fertilization plan are realized, ensuring that subsequent fertilization can be carried out at the time when the target plant has the highest absorption efficiency, effectively improving nitrogen fertilizer utilization rate, and forming a dynamically optimized fertilization strategy. Attached Figure Description
[0017] Figure 1 This is a flowchart of the method of the present invention for calculating and estimating the total amount of nitrogen to be added based on deviation and nitrogen screening threshold; Figure 2 The present invention is a flowchart of the method for determining the recommended total amount of nitrogen fertilizer to be added based on the estimated total amount of nitrogen to be added and the actual demand, and generating an optimized fertilization plan. Figure 3 This is a system module diagram of the present invention. Detailed Implementation
[0018] To make the technical means, creative features, objectives and effects of this invention easier to understand, the specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.
[0019] Example 1 Please see Figure 1-2As shown, this embodiment provides a plant nitrogen fertilizer recommendation method based on UAV spectral diagnostics. Specifically, it is applied to the dynamic identification and optimized regulation of plant nitrogen demand within the UAV flight area. This method achieves intelligent identification, precise regulation, and dynamic feedback optimization of the nitrogen demand of target plants through a closed-loop approach of continuous data acquisition, calculation, correction, and optimization. The method includes the following steps: Target identification and data acquisition: Geographic boundary information of the UAV flight area designated for the target plant is obtained. Within the UAV flight area, to ensure comprehensive and uniform data acquisition and avoid distortion of analysis results due to sampling bias, an image acquisition path covering the entire area is planned. This path has multiple equally spaced image acquisition points. The UAV is instructed to fly along the image acquisition path, and upon reaching each image acquisition point, one or more high-resolution spectral images are acquired using an onboard spectral imaging device, including an imaging spectrometer covering multiple bands such as visible light and near-infrared. The acquired spectral images are processed using an image processing procedure based on pixel spectral values, texture, and spatial proximity relationships. This procedure employs a set of algorithms to automatically identify and extract plant canopy information from the spectral images. The procedure sequentially performs the following steps: preliminary segmentation based on pixel spectral reflectance differences to distinguish vegetation from the background; applying texture analysis algorithms, such as those based on gray-level co-occurrence matrices, to further refine the vegetation area; and using spatial proximity clustering or edge detection algorithms, such as Canny edge detection. The outline boundaries of individual plant canopies were identified and delineated. Combined with spectral feature analysis techniques, such as calculating the Normalized Difference Vegetation Index (NDVI) and red-edge position parameters—spectral indicators related to plant nitrogen content—plant information within the image acquisition area was identified and extracted. This plant information specifically included: a unique plant ID assigned to each plant based on its geographic coordinates; plant distribution density derived by counting the number of identified plant canopies within a unit geographic area; and the precise plant distribution location determined by the UAV's own GPS data and image georegistration information. After acquisition, to facilitate regional management and batch analysis, the plant IDs in the plant information were further refined according to… The geographical coordinates are sorted in ascending or descending order along a certain dimension, such as from north to south or from west to east. Based on this sorting result, a group of target plants that are geographically adjacent and have consecutive numbers are defined as the same batch. This allows for the application of a unified analysis standard to the entire batch, improving processing efficiency. Correspondingly, the positions where plant numbers are discontinuous in the sorting result are determined as batch boundaries. In the plant number sequence sorted by geographical coordinates, logical markers are used to divide different plant management batches. These boundaries typically correspond to the natural boundaries of fields, planting rows, or the boundaries of different crop varieties. Based on the batch boundaries, the start and end ranges of each batch are determined, laying the foundation for subsequent batch data processing.
[0020] Furthermore, nitrogen absorption and curve generation are evaluated. To establish a benchmark that accurately reflects the dynamics of plant nitrogen absorption, and to construct an initial state description that conforms to the interaction between plant physiology and the environment, historical light intensity data, effective light-collecting area data, and environmental temperature data of the target plant during its growth cycle are obtained. The effective light-collecting area data is determined as follows: the projected area is calculated based on the plant canopy outline identified in "Target Determination and Data Acquisition"; the leaf area index is calculated using reflectance data of a specific spectral band; and the canopy projected area and leaf area index are combined using a preset calculation formula to obtain a value that can more accurately reflect the plant's photosynthetic potential than the pure canopy area. The light intensity data and the effective light-collecting area data are used to calculate the first nitrogen absorption efficiency through a process that simulates photosynthetic efficiency. For example, the light intensity value is multiplied by the effective light-collecting area value, and then multiplied by a preset coefficient that represents the efficiency of light energy conversion to biomass. This process is used to calculate the contribution of photosynthesis to the rate of nitrogen absorption in plants. Temperature data is used to simulate the impact of temperature on metabolic rate. For example, based on the temperature value, the corresponding metabolic influence coefficient is retrieved from a pre-set lookup table that stores the correspondence between different temperatures and metabolic rates. The second nitrogen absorption efficiency is then calculated, which is used as a calculation index to simulate the impact of environmental temperature on plant metabolism and nitrogen absorption rate. By weighting the first and second nitrogen absorption efficiencies, the weighting coefficients can be preset according to the sensitivity of different plant varieties or growth stages to light and temperature. This yields an average nitrogen absorption efficiency that comprehensively reflects the impact of major environmental factors, which is then determined as the overall nitrogen absorption efficiency. This is a quantitative index that comprehensively reflects the overall impact of major environmental factors such as light and temperature on plant nitrogen absorption. Based on overall nitrogen uptake efficiency and combined with a pre-defined set of plant growth baseline data describing the standard nitrogen accumulation of a plant variety at different growth stages, the plant growth baseline data is a standardized dataset showing the nitrogen content of a specific plant variety under ideal, stress-free environmental conditions as it changes with growth stages. This data is typically in tabular or function curve form. For each key growth stage, such as seedling, jointing, and flowering, a standard nitrogen accumulation or concentration value is defined, generating an initial nitrogen uptake curve that can predict the nitrogen accumulation process under ideal conditions. This initial nitrogen uptake curve is generated based on plant growth models and current environmental data, and is used to predict the nitrogen accumulation process of the target plant under ideal conditions. This curve uses time as the independent variable and the predicted plant nitrogen accumulation as the dependent variable; its shape is jointly determined by the overall nitrogen uptake efficiency and the plant growth baseline data. Historical nitrogen content data of the target plant is acquired, and its current nitrogen content data is acquired simultaneously. For example, the nitrogen nutrient index is calculated by analyzing the current spectral image. The deviation between the trend of historical nitrogen content data and the current nitrogen content data is calculated to assess the difference between the plant's actual absorption and the initial prediction. To promptly identify and address potential problems affecting plant health, the deviation is compared with a critical value exceeding the normal physiological fluctuation range, derived from historical data statistics. This is a preset nitrogen screening threshold, used to determine whether the plant's nitrogen absorption status is abnormal. This threshold, derived from extensive historical data statistical analysis, represents the upper limit of the normal physiological fluctuation range. If the deviation is greater than or equal to the nitrogen screening threshold, the nitrogen absorption status of the target plant is determined to be abnormal. The system executes a pre-defined anomaly handling process on the target plant. For example, it marks the plant's unique plant number and geographic coordinates as high-risk and generates an alarm message to prompt manual verification. If the deviation is less than the nitrogen screening threshold, the nitrogen absorption status is determined to be normal. The slope or intercept of the initial nitrogen absorption curve is adjusted according to the magnitude and direction of the relationship between historical nitrogen content data and current nitrogen content data to generate a corrected nitrogen absorption curve that is closer to the actual growth status. Based on the initial nitrogen absorption curve, after adjustment according to the actual nitrogen content data of the plant, a dynamic nitrogen absorption curve that more accurately reflects its true growth status can be generated. To further improve the accuracy of predictions, the growth stages of the target plant are divided into multiple consecutive time periods. The nitrogen absorption rate based on the modified nitrogen absorption curve is calculated for each time period. It is determined whether the nitrogen absorption rate calculated based on the modified nitrogen absorption curve shows a continuous downward trend or a trend towards stabilization over at least two consecutive time periods. If so, it indicates that the plant has passed through the rapid growth period and entered the stage of stable nutrient absorption and accumulation. The current time period and its subsequent time periods are then defined as the stable absorption region. If not, the current time period is defined as the unstable absorption region. Based on the modified nitrogen absorption curve and with special reference to the data trends within the stable absorption region, a reference nitrogen absorption curve is generated to guide subsequent fertilization decisions, so as to make more reliable predictions of the nitrogen demand of the plant in the middle and later stages.
[0021] Further, determine the fertilization requirements and total amount calculation; after obtaining the reference nitrogen absorption curve, accurately calculate the total amount of nitrogen to be added, which is the total amount of pure nitrogen element that needs to be supplemented to the target plant. Obtain an initial nitrogen absorption upper limit representing the physiological limit of the plant variety per unit time, as an important constraint condition for calculating the amount of fertilizer, and obtain the target nitrogen content value representing the ideal growth state as nitrogen demand information. A benchmark factor is constructed through a preset calculation procedure, such as subtracting the current nitrogen content data from the target nitrogen content value representing the ideal growth state to obtain the nitrogen deficit, and then weighting the nitrogen deficit and the initial nitrogen absorption upper limit to comprehensively consider the current nitrogen deficiency degree and the plant's maximum absorption capacity. Based on the initial upper limit of nitrogen uptake and the baseline factor, an estimated total amount of nitrogen to be added is calculated and defined as the estimated value, serving as the initial input for subsequent refined calculations. To obtain more accurate results, refined calculations are performed using data from the stable uptake region identified in "Evaluating Nitrogen Uptake and Curve Generation." Based on the nitrogen uptake rate within the stable uptake region and the duration of that region, the nitrogen requirement of the target plant at that stage is calculated and defined as the actual requirement. This more accurately reflects the plant's true nutrient needs at that stage. The difference between the estimated value and the actual requirement is then calculated. The difference between the estimated and actual demand is calculated, and the absolute value of the difference is compared with a preset allowable deviation value. If the absolute value of the difference is less than the allowable deviation value, it indicates that the estimated value is accurate enough and no further iterative correction is needed; the estimated value is directly determined as the total amount of nitrogen to be added. If the absolute value of the difference is greater than or equal to the allowable deviation value, an iterative correction mechanism is initiated. The difference, i.e., the actual demand, is subtracted from the estimated value, and multiplied by a first preset correction multiple used to control the adjustment step size. The first preset correction multiple is a preset value used to control the correction step size in the iterative correction mechanism; this multiple determines the direction of the estimated value towards the actual demand. The convergence speed of actual demand is determined by a smaller value, which ensures convergence stability, while a larger value accelerates the convergence speed. An initial correction is obtained, and the estimated value is added to the initial correction until the convergence condition is met or the preset number of iterations is reached. This yields the initially corrected total nitrogen to be added. To ensure the continuity and stability of the fertilization strategy and avoid drastic fluctuations in fertilizer application due to single calculation errors, the previous optimized total nitrogen to be added is obtained, and the change between the initially corrected total nitrogen to be added and the previous optimized total nitrogen to be added is calculated. This change is defined as the correction difference, ensuring the fertilization strategy... A quantitative indicator calculated with slight continuity is used. If the difference between the previous and subsequent corrections exceeds a preset stability threshold, which limits the maximum allowable change in the recommended fertilizer application rate between two adjacent fertilization cycles, the total amount of nitrogen to be added after the initial correction is adjusted to a limit. This adjustment is done by adding or subtracting the stability threshold from the previous optimized total amount of nitrogen to be added, thus obtaining the current optimized total amount of nitrogen to be added. This current optimized total amount of nitrogen to be added is determined as the total amount of nitrogen to be added. After iterative correction and stability limit adjustment, the final total amount of nitrogen to be added for the current fertilization cycle is determined. This value is the result of a trade-off between the previous total amount of fertilizer and the current initial correction amount. After determining the total amount of nitrogen to be added, to convert it into a specific fertilizer application rate, the nitrogen content of the plant under normal growth standards is obtained and compared with the current nitrogen content data to calculate the nitrogen deficit. Then, the absorption and conversion ratio of the applied nitrogen fertilizer by the standard crop is obtained. Finally, based on the nitrogen deficit and the absorption and conversion ratio, the recommended total amount of commercial fertilizer to be applied by the user is calculated through division.
[0022] Furthermore, the timing of application is determined. To achieve a fertilization strategy of small, frequent, and on-demand application, the recommended total amount of nitrogen fertilizer determined in "Determining Fertilizer Needs and Calculating Total Amount" above is divided into at least two doses. The application time for each dose is determined, and a basic fertilization cycle is set as the time interval, such as 7 days. This cycle forms the basis for calculating the fertilization time points. An offset coefficient for fine-tuning is set, the value of which is dynamically determined by the plant nitrogen absorption effect after the last fertilization in "Recording, Monitoring, and Closed-Loop Correction" below. A positive value indicates an extended cycle, and a negative value indicates a shortened cycle. A dimensionless adjustment factor is used to dynamically fine-tune the fertilization cycle, with the current time... An initial offset, such as 24 hours, is added to ensure sufficient data processing time, which is then used to determine the addition time point for the first dose. The addition time point for each subsequent dose is calculated as follows: based on the addition time point of the previous dose, a time interval is added, and this time interval is adjusted according to the offset coefficient fed back in "Recording, Monitoring and Closed-Loop Correction" below, such as multiplying the time interval by (1 + offset coefficient), thereby determining the next addition time point. The output includes a fertilization plan containing each dose and its corresponding addition time point, which includes the size of each dose into which the recommended total amount of nitrogen fertilizer is applied, as well as the precise addition time point corresponding to each dose.
[0023] Furthermore, recording, monitoring, and closed-loop correction are implemented. To form a self-optimizing closed-loop system, the fertilization effect is continuously tracked and feedback is provided. After applying the corresponding dose of nitrogen fertilizer at each addition time point determined in the fertilization plan, spectral images are continuously collected via drones to record and calculate the changes in nitrogen content of the target plants. Based on the nitrogen content change curve over a period of time after fertilization, the effective nitrogen increment corresponding to a unit weight of nitrogen fertilizer is calculated. The effective nitrogen increment is used as an evaluation index to quantify the effect of applying a unit weight of nitrogen fertilizer. The calculation method is as follows: during an observation period after fertilization, the total increase in plant nitrogen content is divided by the weight of nitrogen fertilizer applied to obtain the effective nitrogen increment in the plant corresponding to a unit weight of nitrogen fertilizer. This quantifies the fertilization effect, accurately records the execution time of nitrogen fertilizer application, and establishes a correlation rule between the effective nitrogen increment and the adjustment of fertilization time, that is, establishing a correlation between fertilization effect (effective nitrogen increment) and... The pre-defined logic and association rules between fertilization time adjustments (offset coefficients) are used to generate the offset coefficients in the "determining the addition time point" section above. If the effective nitrogen increment is higher than the preset expected gain threshold, which represents the minimum effective nitrogen increment that a unit weight of nitrogen fertilizer should bring under ideal conditions, it indicates that the plant absorption efficiency is high, and the next fertilization cycle can be appropriately extended, generating a positive offset coefficient to extend the fertilization interval. Conversely, if it is lower than the expected gain threshold, it indicates that more frequent replenishment is needed, generating a negative offset coefficient. The offset coefficient generated based on this rule will be used to correct the addition time point of the next cycle in the "determining the addition time point" section above, thereby generating the corrected nitrogen fertilizer addition time point and outputting the corrected addition time point to guide the next round of precise fertilization operations, completing the closed-loop control of the entire system.
[0024] The revised nitrogen fertilizer application time point refers to the final execution time obtained by dynamically adjusting the original planned fertilization time point based on feedback from the previous round of fertilization. The calculation method is as follows: based on the original planned time point, the fertilization interval is adjusted using an offset coefficient to achieve more precise on-demand supply.
[0025] Example 2 Please see Figure 3 As shown, this embodiment discloses a plant nitrogen fertilizer recommendation system based on UAV spectral diagnostics. This system, through dynamic monitoring, precise modeling, and closed-loop feedback, intelligently recommends and corrects the total amount, dosage, and timing of nitrogen fertilizer application for target plants, thereby improving nitrogen fertilizer utilization and ensuring healthy crop growth. In practical implementation, this system can be deployed on servers, cloud platforms, workstations, or embedded agricultural IoT devices, and interacts with external devices such as UAVs, weather stations, and soil sensors via wireless or wired networks. The system includes the following modules: The nitrogen status monitoring module is configured to acquire historical and current nitrogen content data of the target plant. Historical nitrogen content data can be retrieved from the system's historical database, which stores the results of previous monitoring. For acquiring current nitrogen content data, the following operations are performed: Within a designated drone flight area for the target plant, such as wheat or corn in a specific farmland plot, one or more efficient image acquisition paths are planned and determined. A drone equipped with a multispectral or hyperspectral camera flies along the image acquisition path, hovering or flying at low speed over multiple preset image acquisition points on the path to acquire high-quality images. The acquired images are processed, such as identifying the target plant in the image through image segmentation algorithms and extracting its spectral information. Then, vegetation indices reflecting the nitrogen nutrition status of the plant are calculated, such as the Normalized Difference Vegetation Index (NDVI) and red-edge position. Based on a pre-established relationship model between spectral indices and nitrogen content, the current nitrogen content data of the target plant is calculated. This is the above image processing and calculation process.
[0026] The fertilization recommendation generation module is the core of the system's decision-making. After receiving historical and current nitrogen content data provided by the nitrogen status monitoring module, it is activated to generate a complete nitrogen fertilizer recommendation plan. Its specific workflow is as follows: Calculate the deviation between historical nitrogen content data and current nitrogen content data. This deviation directly reflects the difference between the target plant's current nitrogen absorption status and historical trends. Compare the deviation with a preset nitrogen screening threshold. If the deviation is greater than or equal to the threshold, the nitrogen absorption status of the target plant is judged to be abnormal, and an alarm can be triggered or a more in-depth diagnostic process can be initiated. Based on the bias, the initial nitrogen absorption curve characterizing the historical nitrogen absorption trend is corrected to generate a corrected nitrogen absorption curve that better reflects the current true growth status of the plant. The process of generating the initial nitrogen absorption curve includes: acquiring environmental data of the target plant in the current growth cycle, specifically including light intensity data, effective light-collecting area data, such as canopy coverage obtained through image analysis, and temperature data; calculating the first nitrogen absorption efficiency based on the light intensity data and effective light-collecting area data; calculating the second nitrogen absorption efficiency based on the temperature data; taking a weighted average or arithmetic average of the two efficiencies to determine the overall nitrogen absorption efficiency; and determining the model parameters used to characterize the initial nitrogen absorption curve based on this overall efficiency. Based on the modified nitrogen absorption curve and the nitrogen requirement information of the target plant (either externally input or preset), such as the total nitrogen requirement corresponding to the target yield, a recommended total amount of nitrogen fertilizer is generated. During the calculation process, the growth stage of the target plant is divided into multiple time periods, and the nitrogen absorption rate in each time period is calculated. If it is determined that the nitrogen absorption rate continues to decline in at least two consecutive time periods, the current time period and its subsequent time periods are defined as the stable absorption region. This helps to more accurately assess the nitrogen requirement of the plant in the later stages. In the specific steps of determining the total amount of nitrogen fertilizer, an optimal strategy is adopted: based on an initial nitrogen absorption upper limit and nitrogen requirement information, an estimated total amount of nitrogen to be added is calculated. Based on data collected within the stable absorption region, the actual nitrogen requirement of the target plant is calculated. The difference between the estimated total nitrogen to be added and the actual requirement is calculated. When this difference is less than a preset allowable deviation value, it indicates that the estimation is relatively accurate, and the system determines the final recommended total nitrogen fertilizer application based on the estimated total nitrogen to be added. Otherwise, when the difference is greater than or equal to the allowable deviation value, the system can directly determine the recommended total nitrogen fertilizer application based on the actual requirement to ensure the practicality of the recommendation. As a more refined alternative, when the difference is large, the system can also execute an iterative correction mechanism: based on the difference, the estimated total nitrogen to be added is corrected to obtain a preliminarily corrected total nitrogen to be added. The module adjusts the amount of nitrogen fertilizer to be added based on the variation between the initial correction value and the previous optimized total amount of nitrogen fertilizer to be added, thus limiting the adjustment to obtain the current optimized total amount of nitrogen fertilizer to be added and avoiding drastic fluctuations in the recommended value. Based on the current optimized total amount of nitrogen fertilizer to be added, the module determines the recommended total amount of nitrogen fertilizer to be applied. The module divides the determined recommended total amount of nitrogen fertilizer to be applied into at least one dose and determines the corresponding addition time point for each dose to form a complete nitrogen fertilizer recommendation scheme. The method for determining the addition time point can be: adding an initial offset to the current time as the addition time point of the first dose; and accumulating a preset time interval based on the addition time point of the previous dose to determine the addition time point of the next dose.
[0027] The fertilization effect evaluation module is responsible for achieving closed-loop feedback and adaptive correction in fertilization management. It starts working after farm managers or automated fertilization equipment have performed at least one nitrogen fertilizer application based on the recommended nitrogen fertilizer plan. It invokes the nitrogen status monitoring module to monitor changes in nitrogen content in the target plants after fertilization. Based on the monitored data, it calculates the effective nitrogen increment per unit of nitrogen fertilizer, such as per kilogram of pure nitrogen. This increment quantifies the actual effect of the fertilization and the plant's absorption efficiency. Based on the calculated effective nitrogen increment, it corrects the previously defined but not yet executed addition time points in the fertilization recommendation generation module to generate corrected addition time points. If the effective nitrogen increment is higher than expected, it indicates strong plant absorption capacity, and subsequent fertilization times can be appropriately advanced; conversely, if the increment is lower than expected, subsequent fertilization can be postponed, or adjustments to the subsequent fertilizer dosage can be recommended.
[0028] Through the collaborative work of the above modules, a complete closed loop is constructed from "data collection - analysis and decision-making - execution - effect evaluation - dynamic correction". This not only provides accurate fertilization plans based on the actual needs of plants, but also continuously optimizes itself based on the real effects after fertilization. It is particularly suitable for the scenario of refined and intelligent nutrient management of crops in modern precision agriculture, which helps to reduce costs, increase efficiency and protect the agricultural ecological environment.
Claims
1. A method for recommending nitrogen fertilizer for plants based on UAV spectral diagnosis, characterized in that, Includes the following steps: Determine the deviation between historical nitrogen content data and current nitrogen content data of the target plant; based on the deviation between historical nitrogen content data and current nitrogen content data of the target plant, correct the initial nitrogen absorption curve characterizing the historical nitrogen absorption trend to generate a corrected nitrogen absorption curve; Based on the corrected nitrogen absorption curve and the nitrogen requirement information of the target plant, a recommended total amount of nitrogen fertilizer is generated, and the recommended total amount of nitrogen fertilizer is divided into at least one dose to determine the corresponding addition time point. After applying nitrogen fertilizer based on the addition time point, the effective nitrogen increment corresponding to one unit of nitrogen fertilizer is calculated based on the monitored changes in nitrogen content of the target plant. The addition time point is corrected based on the effective nitrogen increment to generate the corrected addition time point. The steps for generating the recommended total amount of nitrogen fertilizer to be applied based on the corrected nitrogen absorption curve and the nitrogen requirement information of the target plant specifically include: calculating the estimated total amount of nitrogen to be added based on the initial nitrogen absorption upper limit and nitrogen requirement information; calculating the actual requirement of the target plant based on the data in the stable absorption area; when the difference between the estimated total amount of nitrogen to be added and the actual requirement is less than the preset allowable deviation value, determining the recommended total amount of nitrogen fertilizer to be applied based on the estimated total amount of nitrogen to be added; otherwise, determining the recommended total amount of nitrogen fertilizer to be applied based on the actual requirement.
2. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, The steps for obtaining current nitrogen content data include: determining the image acquisition path within the drone flight area set for the target plant, flying the drone along the acquisition path to acquire images at multiple image acquisition points, and processing the images to identify the target plant and obtain current nitrogen content data.
3. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, Also includes: The deviation between historical nitrogen content data and current nitrogen content data is compared with the preset nitrogen screening threshold; If the deviation is greater than or equal to the nitrogen screening threshold, the nitrogen absorption status of the target plant will be judged as abnormal.
4. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, The steps for revising the initial nitrogen absorption curve, which characterizes the historical nitrogen absorption trend, to generate a revised nitrogen absorption curve include: acquiring light intensity data, effective light-collecting area data, and temperature data of the target plant during its growth cycle; calculating the first nitrogen absorption efficiency based on the light intensity data and the effective light-collecting area data; calculating the second nitrogen absorption efficiency based on the temperature data; determining the average of the first nitrogen absorption efficiency and the second nitrogen absorption efficiency as the overall nitrogen absorption efficiency; and determining the model parameters used to characterize the initial nitrogen absorption curve based on the overall nitrogen absorption efficiency.
5. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, The step of generating the recommended total amount of nitrogen fertilizer based on the modified nitrogen absorption curve and the nitrogen requirement information of the target plant also includes: dividing the growth stage of the target plant into multiple time periods and calculating the nitrogen absorption rate in each time period. If it is determined that the nitrogen absorption rate continues to decline in at least two consecutive time periods, the current time period and its subsequent time periods are defined as the stable absorption region.
6. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, When the difference between the estimated total amount of nitrogen to be added and the actual demand is greater than or equal to the allowable deviation value, the step of determining the recommended total amount of nitrogen fertilizer to be applied based on the actual demand is replaced by: correcting the estimated total amount of nitrogen to be added based on the difference between the estimated total amount of nitrogen to be added and the actual demand to obtain a preliminary corrected total amount of nitrogen to be added; adjusting the preliminary corrected total amount of nitrogen to be added based on the change between the preliminary corrected total amount of nitrogen to be added and the previously optimized total amount of nitrogen to be added to obtain the current optimized total amount of nitrogen to be added; and determining the recommended total amount of nitrogen fertilizer to be applied based on the current optimized total amount of nitrogen to be added.
7. The method for recommending plant nitrogen fertilizer based on UAV spectral diagnosis according to claim 1, characterized in that, The steps for determining the addition time point include: adding an initial offset to the current time to determine the addition time point of the first dose; and accumulating a preset time interval based on the addition time point of the previous dose to determine the addition time point of the next dose.
8. A plant nitrogen fertilizer recommendation system based on UAV spectral diagnostics, characterized in that, Includes the following modules: The nitrogen status monitoring module is used to acquire historical and current nitrogen content data of the target plant. The fertilizer recommendation generation module, in response to the data obtained by the nitrogen status monitoring module, is used to determine the deviation between historical nitrogen content data and current nitrogen content data, and generates a nitrogen fertilizer recommendation scheme based on the deviation, including at least one dosage and the corresponding addition time point. The fertilization effect evaluation module is used to calculate the effective nitrogen increment based on the monitored changes in nitrogen content of the target plants after applying nitrogen fertilizer according to the recommended nitrogen fertilizer plan, and to correct the addition time point based on the effective nitrogen increment to generate the corrected addition time point.
9. A plant nitrogen fertilizer recommendation system based on UAV spectral diagnostics according to claim 8, characterized in that, The nitrogen status monitoring module acquires current nitrogen content data by controlling a drone to collect images within the drone flight area of the target plant; and by processing the images to identify the target plant and acquire its current nitrogen content data.
10. A plant nitrogen fertilizer recommendation system based on UAV spectral diagnostics according to claim 8, characterized in that, The fertilizer recommendation generation module is also used to: compare the deviation with the preset nitrogen screening threshold; if the deviation is greater than or equal to the nitrogen screening threshold, the nitrogen absorption status of the target plant is judged as abnormal. When generating nitrogen fertilizer recommendation plans, the fertilizer recommendation generation module is also used to: calculate the estimated total amount of nitrogen to be added and the actual demand. Based on the comparison between the estimated total amount of nitrogen to be added and the actual demand and the preset allowable deviation value, the recommended total amount of nitrogen fertilizer to be applied is determined selectively according to either the estimated total amount of nitrogen to be added or the actual demand.