Iron-nitrogen compound application method and system based on big data analysis

Abstract

The invention discloses an iron-nitrogen compound application method and system based on big data analysis, and belongs to the field of digital agriculture, the method comprises the following steps: collecting multi-source heterogeneous data of a target farmland, sending the collected multi-source heterogeneous data to a cloud, and constructing a multi-source heterogeneous database in the cloud; data in the multi-source heterogeneous database is called through an iron-nitrogen dynamic application model deployed in the cloud to obtain an iron-nitrogen composite application strategy, an optimal decision scheme is issued to a geographic information system, the geographic information system divides a target farmland into a plurality of grid management units, and the grid management units are used for managing the target farmland. And constructing fertilizer application data based on the information of the optimal decision scheme, and issuing the fertilizer application data to a control system of the automatic agricultural machine, or pushing the fertilizer application data to a user through a user terminal as a basis for manual operation. The invention has beneficial effects.

Description

Technical Field

[0001] This invention relates to the field of digital agriculture, and in particular to a method and system for applying iron-nitrogen compound fertilizer based on big data analysis. Background Technology

[0002] Precision agriculture is one of the core directions of modern agricultural development. Its goal is to achieve high yield, high quality, high efficiency, ecological sustainability, and safety of crops through meticulous field management. Variable-rate fertilization, as a key component of precision agriculture, aims to precisely apply fertilizers according to the actual needs of crops at different growth stages and in different spatial locations, thereby maximizing fertilizer utilization and reducing production costs and environmental pollution.

[0003] In existing variable fertilization techniques, a common practice is to use remote sensing equipment such as drones to acquire Normalized Difference Vegetation Index (NDVI) images of crops. NDVI reflects crop growth and chlorophyll content, thus indirectly characterizing its nitrogen requirement. This technique typically divides farmland into different fertilization management zones based on NDVI values ​​and independently calculates the recommended nitrogen application rate for each management zone based on a pre-set empirical model (e.g., a linear relationship between NDVI values ​​and nitrogen requirement). For micronutrients such as iron (Fe), their application decisions are often separated from those for macronutrients such as nitrogen (N). The usual practice is to apply basal fertilizer once before sowing based on regional soil surveys or a single soil test, lacking a dynamic adjustment mechanism coordinated with nitrogen fertilizer application throughout the crop's growing season. Summary of the Invention

[0004] One of the objectives of this invention is to provide a method for iron-nitrogen compound application based on big data analysis, so as to solve the problem of the lack of a dynamic intelligent fertilization mechanism that coordinates with nitrogen fertilizer application in the existing technology.

[0005] This invention is achieved through the following technical solution: a method for iron-nitrogen compound application based on big data analysis, comprising the following steps: S100, collecting multi-source heterogeneous data of the target farmland, sending the collected multi-source heterogeneous data to the cloud, and constructing a multi-source heterogeneous database in the cloud; S200, obtaining an iron-nitrogen compound application strategy by calling data from the multi-source heterogeneous database through an iron-nitrogen dynamic application model deployed in the cloud, wherein the iron-nitrogen dynamic application model includes: a core physiological model for simulating the intrinsic mechanism of crop growth, a Kalman filter device for continuously calibrating the core physiological model based on observation data, and a dynamic optimization application module for obtaining the optimal decision scheme based on the calibrated core physiological model; S300, distributing the optimal decision scheme to a geographic information system, the geographic information system dividing the target farmland into multiple gridded management units, constructing fertilizer application data based on the information of the optimal decision scheme, distributing the fertilizer application data to the control system of automated agricultural machinery, or pushing it to users through user terminals as a basis for manual operation.

[0006] Furthermore, the multi-source heterogeneous data includes: soil physicochemical parameter data, meteorological data, and crop canopy spectral data.

[0007] Furthermore, the iron-nitrogen dynamic application model was constructed through the following steps:

[0008] S210. A core physiological model for simulating the internal physiological state of crop growth and the influence of external environmental conditions. This core physiological model is based on the iron-nitrogen coupling mechanism. By decomposing the crop-soil system into multiple mutually coupled core processes, a biochemical model describing the internal operating law of the crop-soil system is constructed. S220. After constructing the core physiological model, the state of the core physiological model is dynamically calibrated using extended Kalman filtering. By performing a prediction-update loop at each time step, real-time observation data from multiple sensors are integrated into the core model. S230. A dynamic optimization application model is constructed based on a model predictive control algorithm. A rolling optimization strategy is adopted. At each decision moment, the evolution of the system within a short future time window is predicted using the core physiological model. Then, guided by a comprehensive objective function, the fertilization operation sequence that maximizes overall benefits within this time window is calculated. The prediction-optimization-execution loop is repeated by acquiring the latest sensor data.

[0009] Furthermore, the objective function can be expressed as follows:

[0010]

[0011] And subject to the following constraints:

[0012]

[0013] in, This indicates finding the minimum value; The fertilization rate to be optimized (decision variable); For future prediction time domain; This is the cost coefficient for fertilization operations; This is the penalty coefficient for the output gap; Target biomass; For the future moment Predicted biomass; For collaborative absorption reward coefficient; The iron-nitrogen synergistic absorption index; This is the risk penalty coefficient; This is a future risk index.

[0014] Furthermore, the core physiological model is constructed through the following steps: S211. Based on Liebig's law of minimum quantity, the growth rate of crops is modeled as the result of multiple limiting factors. At the same time, environmental factors such as temperature, light, and water are used as regulatory factors to correct the theoretical maximum growth potential, thus constructing a mathematical description of the dynamic changes in crop biomass; S212. Based on the dynamic changes in crop biomass, in order to quantify the synergistic effect of iron and nitrogen, the kinetic process of crop absorption is further constructed. By introducing a synergistic function, the positive influence of the concentration of different nutrients in the plant on the absorption efficiency of other elements is described; S213. Taking into account the absorption and consumption of plants, the replenishment of artificial fertilizers, and the replenishment caused by leaching losses and organic matter mineralization due to environmental factors, the dynamic changes in the concentration of nutrients in the soil are modeled, thereby constructing a complete material cycle.

[0015] Furthermore, the dynamic changes in crop biomass can be expressed by the following formula:

[0016] in, For crop biomass; For time; This is the rate of change in biomass, i.e., the crop growth rate; This is a function of temperature influence. This is the actual temperature; Let be the function of influence of light radiation; The intensity of light radiation; The function representing the influence of soil moisture; Soil moisture content; The function is for finding the minimum value; This is the response function of iron nutrition to growth; This refers to the iron concentration within the plant. This is the response function of nitrogen nutrition to growth; This represents the nitrogen concentration within the plant.

[0017] Furthermore, the iron concentration within the plant ( The rate of change of ) can be expressed by the following formula:

[0018] in, This represents the rate of change in iron concentration within the plant; The basic iron absorption rate; This is a co-promoting function of nitrogen for iron absorption; This refers to the nitrogen concentration within the plant. This represents the basic loss rate of iron. This represents the current iron concentration within the plant. Standardized factor for losses related to biomass; This refers to crop biomass.

[0019] Similarly, nitrogen concentration in plants ( The rate of change of ) can be expressed by the following formula:

[0020] in, The rate of change of nitrogen concentration within the plant; The basic nitrogen uptake rate; This is a co-promoting function of iron for nitrogen absorption; This refers to the iron concentration within the plant. This represents the basic nitrogen loss rate. This represents the current nitrogen concentration within the plant. Standardized factor for losses related to biomass; This refers to crop biomass.

[0021] Furthermore, soil iron concentration ( ) and nitrogen concentration ( The dynamic changes of ) can be represented by the following system of equations:

[0022] ,

[0023] in, and These represent the rates of change in soil iron and nitrogen concentrations, respectively. The conversion coefficient is related to soil density; This represents the root absorption area. and These represent the net absorption rates of iron and nitrogen concentrations within the plant, respectively. and Let t be the amount of artificial fertilizer input at time t; and This refers to nutrient leaching loss. and This is a supplementary item for organic matter mineralization; Soil moisture; rain is rainfall; OM is soil organic matter content; For temperature.

[0024] Further, S221, predict the state at the current moment based on the state at the previous moment; S222, compare this prediction result with the actual measurement result of the sensor and calculate the error between the two; S223, according to the magnitude of the error and the degree of uncertainty of the system, perform weighted correction on the predicted state of the model to obtain an optimal estimate.

[0025] Furthermore, the state update process is expressed by the following equation:

[0026] ,

[0027] in, This is the system's state vector, which contains all the core variables that need to be tracked in the model; The optimal state estimation vector at time t; This is a vector of predicted values ​​for the current time based on the state at the previous time step; This is the actual observation vector from the sensor at time t, which contains the measurement data directly obtained from the sensor; The observation matrix acts as a mapping, connecting the state variables within the model with observable measurements. K is the Kalman gain matrix, which is the core of state updates. By comprehensively considering the uncertainty of prediction (process noise) and the uncertainty of sensor measurement (observation noise), it intelligently determines the extent to which new measurement data should be trusted. If the sensor is very accurate, the weight of K is large, and the model state will be pulled more towards the measured value; otherwise, the model's prediction will be trusted more.

[0028] Furthermore, the iron-nitrogen dynamic application model also includes: S240, an assessment module for quantifying future environmental risks, used to achieve forward-looking decision-making. It quantifies risks by constructing a comprehensive risk index, taking predicted temperature anomalies, predicted soil moisture deviations, and the probability of extreme weather events occurring in the future time period as inputs and assigning them different weights. Through weighted summation, it integrates multiple environmental factors that threaten crop nutrient absorption, thereby obtaining a comprehensive risk index that reflects the degree of impact of different risk factors on crop yield. This comprehensive risk index is used as a penalty term in optimization decision-making.

[0029] Furthermore, the comprehensive risk index is calculated using the following formula:

[0030] ,

[0031] in, The comprehensive risk index for future assessment at time t; These are weighting coefficients for temperature, humidity, and extreme weather risks, used to balance the importance of different types of risks; This represents the absolute value of the predicted future temperature anomaly deviation. This represents the absolute value of the predicted future soil moisture anomaly deviation; This represents the probability of extreme weather events occurring in the future.

[0032] Furthermore, the optimal decision-making scheme is a high-value variable data set. This data is used in the geographic information system to divide the entire plot to be fertilized into a prescription map composed of multiple gridded management units. For each management unit, the core content of the prescription map is generated in a refined manner based on its independent model state and the optimal decision-making scheme calculated by the MPC algorithm.

[0033] Furthermore, the prescription chart includes: precise execution instructions for the current cycle: these are the best decisions that automated agricultural machinery or operators need to strictly follow, after comprehensive optimization; a forward-looking recommended application time window: providing a dynamic application plan within the future prediction time domain, showing the recommended fertilizer amount and optimal application time in the form of a timeline or list, forming a recommended application time window. At the same time, the system will clearly mark that this future plan is tentatively optimal, because it will be updated and optimized again at the beginning of the next decision cycle based on the latest sensor data; and intelligent fertilizer type suggestions: the prescription chart also transforms the abstract element demand from the model output into specific fertilizer type suggestions.

[0034] Another aspect of the present invention provides an iron-nitrogen compound application system based on big data analysis, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the multimodal data-driven integrated environmental management method for agricultural product sources as described above.

[0035] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0036] 1. This invention quantifies the synergistic relationship between iron and nitrogen in fertilization decision-making by constructing an iron-nitrogen physiological coupling sub-model. This makes the generated fertilization plan no longer a simple superposition of the two nutrients, but ensures that their ratio and dosage can maximize the absorption and utilization of each other, thereby improving the comprehensive utilization rate of fertilizer and achieving the effect of synergistic yield increase.

[0037] 2. This invention integrates real-time data from multiple sources, such as soil, meteorology, and remote sensing, and uses this data to drive a dynamic model to simulate the crop growth process. This enables fertilization decisions to respond in real time to the actual growth status of the crop and constantly changing environmental conditions, realizing the transformation from static snapshots to dynamic simulation.

[0038] 3. This invention incorporates future weather forecast data into a multi-objective optimization function, achieving proactive risk correction. By delaying nitrogen application before forecasting heavy rain or recommending the use of more easily absorbed fertilizer forms before forecasting drought, proactive risk avoidance is achieved. This proactive risk avoidance mechanism can significantly reduce the risk of yield reduction and fertilizer loss caused by extreme weather events, ensuring the stability and sustainability of agricultural production. Attached Figure Description

[0039] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0040] Figure 1 This is a flowchart of the method provided in Embodiment 1 of the present invention.

[0041] Figure 2 The flowchart is for the iron-nitrogen dynamic application model provided in Embodiment 1 of the present invention.

[0042] Figure 3 This is a time series diagram of the iron-nitrogen dynamic application model provided in Embodiment 1 of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0044] Example 1

[0045] This embodiment discloses a method for applying iron-nitrogen composite fertilizer based on big data analysis. Figure 1 The overall method flowchart of this embodiment is shown. As can be seen from the figure, this embodiment includes the following steps:

[0046] Step 1: Collect multi-source heterogeneous data of the target field and send the collected multi-source heterogeneous data to the cloud to build a multi-source heterogeneous database in the cloud.

[0047] Specifically, multi-source heterogeneous data can include: soil physicochemical parameter data: multiple IoT soil sensors are deployed in a grid pattern in the plot to be fertilized; these sensors can collect and upload data such as soil volumetric water content, temperature, pH value, and electrical conductivity (EC) in real time. Soil pH and EC values ​​are particularly important, as these two values ​​directly affect the availability of iron and nitrogen in the soil and are key inputs for subsequent analysis.

[0048] Meteorological data: Accurate meteorological data for the location of the land parcel can be obtained by calling publicly available commercial meteorological service APIs. This includes historical meteorological data (such as temperature, precipitation, and sunshine hours) over a past period, as well as weather forecast data for a preset future period.

[0049] Crop canopy spectral data: Periodically (e.g., every 5-7 days) using drones or satellites equipped with multispectral or hyperspectral cameras, remote sensing data is collected on the plots. This spectral data can be used to assess the real-time growth and nutritional status of crops.

[0050] Step 2: Deploy an iron-nitrogen dynamic application model in the cloud. This model can call data from a multi-source heterogeneous database for analysis to obtain an iron-nitrogen composite application strategy.

[0051] Specifically, the iron-nitrogen dynamic application model in this implementation simulates the intrinsic mechanisms of crop growth through a core physiological model. Then, Kalman filtering is used to continuously calibrate the core model using observational data. Finally, based on the calibrated core model, a forward-looking decision engine is used to select the optimal decision. This transforms multi-source, heterogeneous real-time data streams and future forecast information into spatiotemporally continuous decision instructions with forward-looking, robust, and economically optimal characteristics. It intelligently finds a dynamic balance among multiple conflicting objectives such as yield, cost, efficiency, and risk, thereby maximizing agricultural productivity and ensuring sustainable resource utilization.

[0052] Specifically Figure 2 The flowchart of the dynamic iron-nitrogen application model in this embodiment is shown. As can be seen from the flowchart, the dynamic iron-nitrogen application model in this embodiment may include the following:

[0053] 1) Construct a core physiological model that can characterize the complex biochemical processes inside crops and external environmental factors to simulate the dynamic interaction between crop growth and iron-nitrogen nutrients.

[0054] Generally, crop growth is not determined by a single factor; it is often driven by both the crop's internal physiological state (such as nutrient levels) and external environmental conditions (such as temperature, light, and water). In particular, there is a complex synergistic relationship between iron (Fe) and nitrogen (N), two key nutrients in crop growth: sufficient nitrogen promotes iron absorption, while iron is a component of key enzymes in nitrogen metabolism. Therefore, this complex dynamic reaction process can be simulated by constructing an iron-nitrogen coupling mechanism.

[0055] This embodiment uses a physiological core model based on the iron-nitrogen coupling mechanism. By decomposing the entire crop-soil system into several mutually coupled core processes, a biochemical model describing the intrinsic operational laws of the crop-soil system is constructed. Specifically, it includes the following:

[0056] First, based on Liebig's law of minimum, the growth rate of crops is modeled as being influenced by multiple limiting factors, meaning that the growth rate of crops is determined by the most scarce resource. Simultaneously, environmental factors such as temperature, light, and water are used as moderating factors to correct for the theoretical maximum growth potential, thereby constructing a mathematical description of the dynamic changes in crop biomass.

[0057] For example, in this embodiment, the dynamic changes in crop biomass can be represented by the following formula:

[0058]

[0059] in, For crop biomass; For time; This is the rate of change in biomass, i.e., the crop growth rate; This is a function of temperature influence. This is the actual temperature; Let be the function of influence of light radiation; The intensity of light radiation; The function representing the influence of soil moisture; Soil moisture content; The function is for finding the minimum value; This is the response function of iron nutrition to growth; This refers to the iron concentration within the plant. This is the response function of nitrogen nutrition to growth; This represents the nitrogen concentration within the plant.

[0060] The temperature effect function can be expressed as follows:

[0061] ,

[0062] in, For the maximum growth rate, It is a natural constant. For the optimal temperature, This represents the temperature tolerance range of the crop; the function is a Gaussian function, reflecting that there is an optimal temperature for crop growth. When the actual temperature When the growth rate deviates from this optimal value, it will decrease exponentially.

[0063] The influence function of light radiation can be expressed as follows:

[0064] ,

[0065] in, This is the half-saturation constant of the plant's photosynthetic response, which is a saturation curve (Michaelis equation) representing the rate of photosynthesis as a function of light radiation intensity. The amount of light increases with the amount of light, but it tends to saturate when the light reaches a certain level.

[0066] The soil moisture influence function can be expressed as follows:

[0067] ,

[0068] in, The minimum soil moisture content, This represents the optimal humidity. It is a simplified linear model representing the optimal humidity when the soil moisture content is... From the lowest soil moisture content During the process of rising to the optimal humidity, that is, the effect of changes in soil moisture content on growth is linear.

[0069] The response function of iron nutrition to growth can be expressed as follows:

[0070] ,

[0071] The response function of nitrogen nutrition to growth can be expressed as follows:

[0072] ,

[0073] in, The limiting threshold for iron, This represents the nitrogen limiting threshold. Both response functions also employ saturation curves to represent the nutrient concentration within the plant. or The higher the value, the stronger the growth-promoting effect, but a saturation effect also exists. and These are the corresponding half-saturation constants, representing the threshold at which nutrients begin to become limiting factors.

[0074] Secondly, based on the established dynamic changes in crop biomass, in order to quantify the synergistic effect of iron and nitrogen, it is necessary to further construct the kinetic process of crop absorption. For the absorption process of iron and nitrogen, this process is not regarded as an independent event, but rather a synergistic function is introduced to use mathematical language to describe how the concentration of different nutrients in the plant positively affects the absorption efficiency of other elements.

[0075] For example, in this embodiment, the iron concentration in the plant ( The rate of change of ) can be expressed by the following formula:

[0076]

[0077] in, This represents the rate of change in iron concentration within the plant; The basic iron absorption rate; This is a co-promoting function of nitrogen for iron absorption; This refers to the nitrogen concentration within the plant. This represents the basic loss rate of iron. This represents the current iron concentration within the plant. Standardized factor for losses related to biomass; This refers to crop biomass.

[0078] Similarly, nitrogen concentration in plants ( The rate of change of ) can be expressed by the following formula:

[0079]

[0080] in, The rate of change of nitrogen concentration within the plant; The basic nitrogen uptake rate; This is a co-promoting function of iron for nitrogen absorption; This refers to the iron concentration within the plant. This represents the basic nitrogen loss rate. This represents the current nitrogen concentration within the plant. Standardized factor for losses related to biomass; This refers to crop biomass.

[0081] It should be noted that both of the above equations consist of an absorption term and a loss term. The loss term represents the expenditure of iron and nitrogen, respectively, simulating the process by which iron and nitrogen are consumed in metabolism or lost due to plant tissue shedding. The absorption term... and This reflects the absorption rate:

[0082] This describes the basic iron absorption rate;

[0083] This describes the basic nitrogen uptake rate;

[0084] in, For the optimal pH value, This is an empirical formula. This represents the maximum absorption rate of iron. This represents the maximum nitrogen uptake rate. The half-saturation constant of iron in the soil. is the half-saturation constant of soil nitrogen. For soil iron concentration, Soil nitrogen concentration, The soil's pH level. The effects of soil moisture and temperature are considered. The basic iron uptake rate depends on the iron concentration and pH of the soil. A nitrogen co-promoting function shows that the higher the nitrogen concentration within the plant, the higher the iron uptake rate. The basic nitrogen uptake rate depends on soil nitrogen concentration and soil temperature and humidity conditions. An iron co-promoting function is used to represent the promoting effect of iron concentration within the plant on nitrogen uptake.

[0085] The two synergistic promotion functions can be expressed by the following formulas:

[0086] ,

[0087] ,

[0088] in, The coefficient of coordination; This represents the iron concentration threshold, indicating the range within which iron optimizes nitrogen metabolism. The promoting effect of iron concentration on nitrogen uptake in plants is a saturated function, meaning that when the iron concentration falls below the threshold... At this time, nitrogen absorption will be significantly limited.

[0089] Finally, by comprehensively considering the absorption and consumption of plants, the replenishment of artificial fertilizers, and the leaching loss caused by environmental factors such as rainfall and the replenishment brought about by organic matter mineralization, the dynamic changes of nutrient element concentration in the soil are modeled, thereby constructing a complete material cycle.

[0090] For example, in this embodiment, soil iron concentration ( ) and nitrogen concentration ( The dynamic changes of ) can be represented by the following system of equations:

[0091] ,

[0092] in, and These represent the rates of change in soil iron and nitrogen concentrations, respectively. The conversion coefficient is related to soil density; This represents the root absorption area. and These represent the net absorption rates of iron and nitrogen concentrations within the plant, respectively. and Let t be the amount of artificial fertilizer input at time t; and This refers to nutrient leaching loss. and This is a supplementary item for organic matter mineralization; Soil moisture; rain is rainfall; OM is soil organic matter content; For temperature.

[0093] It should be noted that the differential equations shown above follow the principle of mass balance and are applied to soil. This set of equations characterizes the material cycle of the entire crop-soil system. In the first term of the equations, by closely linking the soil model with the plant model and using plant uptake as a key input, fertilization, plant growth, and soil nutrient consumption form a complete causal chain. Furthermore, by introducing the artificial fertilization input at time t, the entire equation set is endowed with decision-making capabilities, and this artificial fertilization input becomes the key to subsequent optimization control. Finally, by combining the terms leaching and mineralization, the complex environmental interactions in open ecosystems are considered, making the entire equation set more realistic and enhancing prediction accuracy.

[0094] 2) After constructing the physiological core model, it is necessary to organically integrate real-time observation data from multiple sensors into the core model, so as to achieve dynamic calibration of the core model's state and ensure that the model can keep in sync with the real world.

[0095] Specifically, the core model is essentially a theory-based simulator, and its predictions will inevitably deviate from reality. The predictions of the core model can be corrected using real-world measurements, such as remote sensing satellite data (e.g., the NDVI vegetation index, which reflects crop growth) and soil sensor data (which directly measures soil nutrient concentration).

[0096] In this embodiment, the state of the core model is dynamically calibrated using an extended Kalman filter (EKF). This is achieved through a prediction-update loop at each time step: first, the state at the current time is predicted based on the state at the previous time step; then, this prediction is compared with the actual measurement results from the sensor, and the error between the two is calculated; finally, the predicted state of the model is weighted and corrected according to the magnitude of the error and the degree of uncertainty in the system, resulting in an optimal estimate that is closer to the true value.

[0097] For example, in this embodiment, the state update process can be expressed by the following formula:

[0098]

[0099] in, This is the system's state vector, containing all the core variables that need to be tracked in the model. ; The optimal state estimation vector at time t; This is a vector of predicted values ​​for the current time based on the state at the previous time step; This is the actual observation vector from the sensor at time t, which contains the measurement data directly obtained from the sensor; The observation matrix acts as a mapping, connecting the state variables within the model with observable measurements. K is the Kalman gain matrix, which is the core of state updates. By comprehensively considering the uncertainty of prediction (process noise) and the uncertainty of sensor measurement (observation noise), it intelligently determines the extent to which new measurement data should be trusted. If the sensor is very accurate, the weight of K is large, and the model state will be pulled more towards the measured value; otherwise, the model's prediction will be trusted more.

[0100] 3) After accurately tracking the current state of crops and soil through the aforementioned two steps, it is necessary to establish an assessment module capable of quantifying future environmental risks to achieve forward-looking intelligent decision-making. An optimal fertilization strategy must consider not only the present but also future uncertainties, such as the possibility that upcoming hot weather may inhibit nutrient absorption, or that a heavy rain may cause significant nutrient loss. Therefore, it is necessary to transform future information such as weather forecasts into a quantifiable risk index to mitigate these risks during decision-making.

[0101] Specifically, this embodiment quantifies risk by constructing a comprehensive risk index. By using the predicted probability of future temperature anomalies, predicted soil moisture deviations, and extreme weather events (such as drought and torrential rain) as inputs, and assigning different weights to each, a weighted summation is used to integrate multiple environmental factors that may threaten crop nutrient absorption in the future. Specific weight values ​​can be set based on the experience of agricultural experts or obtained through historical data regression analysis, reflecting the degree of impact of different risk factors on crop yield. This comprehensive risk index ultimately serves as an important penalty factor in optimization decisions, ensuring that the generated fertilization strategy is robust and safe.

[0102] For example, in this embodiment, the comprehensive risk index can be calculated using the following formula:

[0103]

[0104] in, The comprehensive risk index for future assessment at time t; These are weighting coefficients for temperature, humidity, and extreme weather risks, used to balance the importance of different types of risks; This represents the absolute value of the predicted future temperature anomaly deviation. This represents the absolute value of the predicted future soil moisture anomaly deviation; This represents the probability of extreme weather events occurring in the future. The goal of the above formula is to transform ambiguous, multi-dimensional weather forecast information into a clear, one-dimensional numerical value, making it directly usable by subsequent optimization decision-making models.

[0105] 4) Finally, based on all the information from the preceding steps, a dynamic optimization application module is constructed to calculate the optimal iron-nitrogen combined application strategy. This intelligently resolves the complex issues of when, where, what type of fertilizer, and how much to apply. This application strategy has multiple objectives: maximizing final yield, minimizing fertilizer and labor costs, and mitigating the environmental risks assessed in the previous step.

[0106] Specifically, in this embodiment, a dynamic optimization application model can be constructed based on the Model Predictive Control (MPC) algorithm. MPC does not plan the fertilization scheme for the entire growing season all at once, but rather employs a rolling optimization strategy. At each decision point (e.g., each day), using the aforementioned core model, it predicts the system's evolution over a shorter future time window (e.g., the next 3 days). Then, guided by a comprehensive objective function, it calculates the fertilization operation sequence that maximizes overall benefits within this time window. The objective function comprehensively considers the operating costs of fertilization, the gap with the target yield (yield loss penalty), the synergistic efficiency of iron-nitrogen absorption, and future environmental risks. After calculation, MPC does not execute the entire sequence, but only the first operation in the sequence (i.e., the fertilization decision for the day). On the next day, by acquiring the latest sensor data, it repeats the 'prediction-optimization-execution' cycle, enabling the decision to continuously adapt to real-time environmental changes, thereby ensuring the dynamics and robustness of the entire system.

[0107] For example, in this embodiment, the objective function can be expressed as follows:

[0108]

[0109] And subject to the following constraints:

[0110]

[0111] in, This indicates finding the minimum value; The fertilization rate to be optimized (decision variable); For future prediction time domain; This is the cost coefficient for fertilization operations; This is the penalty coefficient for the output gap; Target biomass; For the future moment Predicted biomass; For collaborative absorption reward coefficient; The iron-nitrogen synergistic absorption index; This is the risk penalty coefficient; This is a future risk index.

[0112] It should be noted that in the above formula, the optimization objective is to predict the time domain. Inside, searching for the optimal iron fertilizer and nitrogen fertilizer Application rate sequence. The first term within the integral. This represents the direct operational cost of fertilization; the second item It is a secondary penalty term used to penalize predicted biomass. With target output The difference between them, coefficient This determines the level of importance attached to production losses; the third item It is a reward item, including the synergy index; This is used to measure the synchronicity and efficiency of iron and nitrogen absorption. By rewarding this, the model is encouraged to find fertilization strategies that maximize the synergistic absorption of both. (Extra-integral) This is the risk penalty term, which integrates the previously calculated comprehensive risk index into the optimization objective to ensure the safety of the decision. The constraints ensure that during the optimization process, the predicted output cannot fall below a certain baseline, and future risks cannot exceed a set maximum threshold. Furthermore, the amount of fertilizer applied cannot be negative. By solving this optimization problem, a precise spatiotemporal fertilization strategy sequence is ultimately output. This guides specific fertilization operations at different times and in different field grid locations. Figure 3 A timeline diagram of the iron-nitrogen dynamic application model in this embodiment is shown.

[0113] Step 3: The obtained iron-nitrogen compound application strategy is distributed from the cloud to the geographic information system. The geographic information system divides the plot to be fertilized into multiple gridded management units and constructs fertilizer application data based on the information of the iron-nitrogen compound application strategy. The application data includes the precise application amount of iron fertilizer and nitrogen fertilizer (unit: kg / mu) for each management unit, the recommended application time window, and the suggested fertilizer type.

[0114] The geographic information system then distributes fertilizer application data to the control system of automated agricultural machinery, or pushes it to farm managers via mobile apps and other user terminals as a basis for manual operation.

[0115] Specifically, the optimal fertilization strategy sequence obtained in step 2 This can be viewed as a highly dynamic and information-rich digital variable data set. In a Geographic Information System (GIS), this data divides the entire plot to be fertilized into a prescription map composed of multiple gridded management units (e.g., 10m x 10m). For each management unit, the core content of the prescription map is the optimal spatiotemporal fertilization strategy sequence calculated by the MPC algorithm based on its independent model state. It is generated in a refined manner. Its detailed content may include:

[0116] Precise execution instructions for the current cycle: The primary information in the prescription chart is the immediate execution instruction, which clearly indicates the precise amounts of iron and nitrogen fertilizers to be applied in the current decision-making cycle, directly corresponding to the first element in the optimal strategy sequence. It is the best current decision that is fully optimized and needs to be strictly followed by automated agricultural machinery or operators.

[0117] Proactive recommended application time window: Unlike static fertilization recommendations, this prescription map utilizes the later part of the strategy sequence. It provides a dynamic application plan for a future forecast period (e.g., the next 2 to 7 days), clearly displaying the recommended fertilizer application amounts and optimal application times for the following days in the form of a timeline or list, forming a recommended application time window. Simultaneously, the system explicitly marks this future plan as tentatively optimal, as it will be continuously updated and optimized again at the start of the next decision-making cycle based on the latest sensor data. This provides users with valuable information for production planning and resource allocation.

[0118] Intelligent fertilizer type recommendations: The prescription diagram can also display the required quantity of abstract elements output by the model ( , This information is then transformed into specific, actionable fertilizer type recommendations. A post-optimization module can obtain real-time data on the nutrient content and solubility of different types of fertilizers (e.g., urea, compound fertilizer, ferrous sulfate, EDDHA-Fe chelated iron), and link these data to future climate risks (e.g., automatically increasing the recommended weight of easily absorbed chelated iron fertilizers when MPC predicts drought risk), to calculate fertilizer combinations that meet the target element requirements.

[0119] Ultimately, this detailed, dynamic, and static prescription map can be directly loaded into the control system of automated agricultural machinery supporting variable fertilization (such as variable fertilizer applicators and plant protection drones), enabling fully automated operation. Furthermore, a complete view including forward-looking plans can be pushed to farm managers via mobile apps and other user terminals to assist them in making more scientific agricultural plans. In each new decision-making cycle, the system will regenerate a completely new prescription map based on the latest field data feedback, ensuring a closed-loop, adaptive, and continuously optimized fertilization management process.

[0120] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for applying iron-nitrogen composite fertilizer based on big data analysis, characterized in that, The combined application method includes: S100: Collect multi-source heterogeneous data of the target farmland, send the collected multi-source heterogeneous data to the cloud, and build a multi-source heterogeneous database in the cloud. S200 uses a dynamic iron-nitrogen application model deployed in the cloud to retrieve data from a multi-source heterogeneous database to obtain an iron-nitrogen composite application strategy. The iron-nitrogen dynamic application model includes a core physiological model for simulating the intrinsic mechanisms of crop growth. Kalman filtering devices used to continuously calibrate the core physiological model based on observational data, and, Based on the calibrated core physiological model, a dynamic optimization application module for the optimal decision-making scheme is obtained; S300: The optimal decision-making plan is sent to the Geographic Information System (GIS). The GIS divides the target farmland into multiple gridded management units and constructs fertilizer application data based on the information from the optimal decision-making plan. Fertilizer application data is sent to the control system of automated agricultural machinery or pushed to users through user terminals as a basis for manual operation.

2. The iron-nitrogen combined application method based on big data analysis according to claim 1, characterized in that, The multi-source heterogeneous data includes: soil physicochemical parameter data, meteorological data, and crop canopy spectral data.

3. The iron-nitrogen combined application method based on big data analysis according to claim 1, characterized in that, The iron-nitrogen dynamic application model was constructed through the following steps: S210. A core physiological model for simulating the internal physiological state of crop growth and the influence of external environmental conditions. This core physiological model is based on the iron-nitrogen coupling mechanism. By decomposing the crop-soil system into multiple mutually coupled core processes, a biochemical model describing the internal operating law of the crop-soil system is constructed. S220. After constructing the core physiological model, the state of the core physiological model is dynamically calibrated by using extended Kalman filtering. By performing a prediction-update loop at each time step, real-time observation data from multiple sensors are integrated into the core model. S230. A dynamic optimization application model is constructed based on the model predictive control algorithm. A rolling optimization strategy is adopted, using the core physiological model at each decision moment to predict the system's evolution within a short future time window. Then, guided by a comprehensive objective function, the fertilization operation sequence that maximizes the overall benefit within this time window is calculated. And by acquiring the latest sensor data, the prediction-optimization-execution cycle is repeated.

4. The iron-nitrogen combined application method based on big data analysis according to claim 3, characterized in that, The core physiological model was constructed through the following steps: S211. Based on Liebig's law of minimum, the growth rate of crops is modeled as being influenced by multiple limiting factors. Meanwhile, Environmental factors such as temperature, light, and water are used as regulatory factors to correct the theoretical maximum growth potential and to construct a mathematical description of the dynamic changes in crop biomass. S212. Based on the dynamic changes in crop biomass, in order to quantify the synergistic effect of iron and nitrogen, the dynamic process of crop absorption is further constructed. By introducing a synergistic function, the positive influence of the concentration of different nutrients in the plant on the absorption efficiency of other elements is described. S213. Taking into account the absorption and consumption of plants, the replenishment of artificial fertilizers, and the leaching loss caused by environmental factors and the replenishment brought about by organic matter mineralization, the dynamic changes of nutrient element concentration in the soil are modeled to construct a complete material cycle.

5. The iron-nitrogen combined application method based on big data analysis according to claim 3, characterized in that, The extended Kalman filter enables the dynamic calibration of the core physiological model's state update process, which includes: S221. Predict the state at the current moment based on the state at the previous moment; S222. Compare this prediction result with the actual measurement result of the sensor, and calculate the error between the two. S223. Based on the magnitude of the error and the degree of uncertainty of the system, the predicted state of the model is weighted and corrected to obtain an optimal estimate.

6. The iron-nitrogen combined application method based on big data analysis according to claim 3, characterized in that, The iron-nitrogen dynamic application model also includes: S240, an assessment module for quantifying future environmental risks, used to achieve forward-looking decision-making, and quantifying risks by constructing a comprehensive risk index. By using predicted temperature anomalies, predicted soil moisture deviations, and the probability of extreme weather events in the future as inputs and assigning different weights, a weighted summation method is used to integrate multiple environmental factors that threaten crop nutrient absorption, thereby obtaining a comprehensive risk index that reflects the degree of impact of different risk factors on crop yield. The resulting comprehensive risk index will be used as a penalty factor when making optimization decisions.

7. The iron-nitrogen combined application method based on big data analysis according to claim 6, characterized in that, The comprehensive risk index is calculated using the following formula: ， in, The comprehensive risk index for future assessment at time t; These are weighting coefficients for temperature, humidity, and extreme weather risks, used to balance the importance of different types of risks; This represents the absolute value of the predicted future temperature anomaly deviation. This represents the absolute value of the predicted future soil moisture anomaly deviation; This represents the probability of extreme weather events occurring in the future.

8. The iron-nitrogen combined application method based on big data analysis according to claim 1, characterized in that, The optimal decision-making scheme is a high-value variable data set, which, in a geographic information system, divides the entire plot to be fertilized into a prescription map composed of multiple gridded management units. For each management unit, the core content of the prescription map is generated by refining the optimal decision scheme calculated by the MPC algorithm based on its independent model state.

9. The iron-nitrogen combined application method based on big data analysis according to claim 1, characterized in that, The prescription diagram includes: The precise execution instructions for the current cycle are the best decisions that have been fully optimized and must be strictly followed by automated agricultural machinery or operators. A forward-looking recommended application window: This provides a dynamic application plan within a future forecast timeframe, displaying recommended application amounts and optimal application times in a timeline or list format, forming a recommended application window. The system will clearly indicate that the future plan is tentatively optimal, because it will be updated and optimized again at the beginning of the next decision-making cycle based on the latest sensor data. Intelligent fertilizer type recommendations: The prescription map also transforms the abstract element requirements output by the model into specific fertilizer type recommendations.

10. An iron-nitrogen combined application system based on big data analysis, characterized in that, The combined application system includes: processor; The memory stores a computer program that, when executed by a processor, implements the iron-nitrogen composite application method based on big data analysis as described in any one of claims 1 to 9.

Images