Agricultural crop growth stage fertilizer intelligent regulation and control method based on multi-source factors

Abstract

The invention relates to the technical field of intelligent regulation and control, in particular to an agricultural crop growth stage fertilizer intelligent regulation and control method based on multi-source factors. The method comprises the following steps: obtaining corresponding growth state data, soil environment data, meteorological data and fertilizer application historical data, carrying out multi-source heterogeneous data fusion and standardization processing, and constructing an agricultural crop four-dimensional regulation and control matrix; constructing a crop growth-fertilizer response model based on the agricultural crop four-dimensional regulation and control matrix, and simulating to generate a fertilizer regulation and control simulation data set; performing regulation and control coupling analysis based on the fertilizer regulation and control simulation data set to determine an initial fertilizer application scheme set corresponding to each growth stage; and designing a dynamic adjustment mechanism according to the initial fertilizer application scheme set, and performing comparative analysis and fertilizer scheme re-optimization to output a corresponding optimal fertilizer regulation and control scheme, thereby realizing dynamic visualization and execution of agricultural crop fertilizer regulation and control. The fertilizer utilization efficiency in the growth stage of agricultural crops can be improved to the maximum extent.

Description

Technical Field

[0001] This invention relates to the field of intelligent regulation technology, and in particular to an intelligent regulation method for fertilizers during the growth stages of agricultural crops based on multiple factors. Background Technology

[0002] Crop growth is dynamically influenced by multiple factors, including its physiological stages, soil fertility, and meteorological conditions. Traditional fertilization methods relying on experience and fixed-cycle fertilization are no longer adequate to meet the differentiated nutrient needs of crops at different growth stages, nor can they respond in real-time to changes in soil nutrients and meteorological fluctuations (such as nutrient loss due to rainfall and accelerated nutrient consumption due to high temperatures). This easily leads to fertilizer waste, soil compaction, or nutrient imbalances in crops, affecting yield and quality. Intelligent fertilizer regulation technology based on the integration of multiple factors at different crop growth stages, with its unique advantages of integrating multi-dimensional data and dynamically matching fertilization plans, is gradually replacing the traditional extensive fertilization model. It has become an important technological means to promote precision agriculture, improve fertilizer utilization efficiency, and protect the agricultural ecological environment.

[0003] Currently, several methods for regulating fertilizer use in agricultural crops have been proposed. These methods typically involve collecting basic soil nutrient data or single crop growth indicators, combining them with fixed fertilizer formulas to develop fertilization plans, and then executing fertilization operations according to a preset cycle. However, existing methods lack sufficient synergistic analysis and dynamic adaptation of multi-source factors. They are prone to disconnect between fertilization plans and actual needs due to factors such as not considering the impact of meteorological conditions on nutrient absorption and neglecting the real-time correlation between crop growth stages and soil fertility. Furthermore, they lack dynamic adjustment and simulation verification mechanisms, and can only be executed according to fixed plans, making it difficult to optimize fertilization parameters based on the real-time growth status of crops, thus reducing the accuracy of fertilizer regulation. Summary of the Invention

[0004] Therefore, it is necessary for the present invention to provide a method for intelligent regulation of fertilizers during the growth stages of agricultural crops based on multiple factors, in order to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, a method for intelligent fertilizer regulation during the growth stages of agricultural crops based on multiple factors includes the following steps: Step S1: Obtain the growth status data, soil environment data, meteorological data and fertilizer application history data of agricultural crops, and perform multi-source heterogeneous data fusion and standardization processing to generate standardized crop growth characteristic datasets, soil environment characteristic datasets, meteorological impact datasets and fertilizer application characteristic datasets. Step S2: Extract the physiological index change patterns of crops at different growth stages based on crop growth characteristic datasets, and explore the correlation between soil fertility and crop nutrient absorption efficiency by combining soil environmental characteristic datasets. At the same time, introduce meteorological impact datasets and fertilizer application characteristic datasets to construct a four-dimensional regulation matrix for agricultural crops. Step S3: Construct a crop growth-fertilizer response model based on the four-dimensional regulation matrix of agricultural crops, and simulate the crop growth process under different fertilizer application schemes based on the crop growth-fertilizer response model to generate a fertilizer regulation simulation dataset; perform regulation coupling analysis based on the fertilizer regulation simulation dataset and the crop growth characteristic dataset to determine the initial fertilizer application scheme set corresponding to each growth stage that meets the crop nutrient requirements and avoids soil nutrient imbalance. Step S4: Design a dynamic adjustment mechanism based on the initial fertilizer application scheme set to compare and analyze the real-time crop growth status data with the fertilizer regulation simulation dataset to generate fertilizer regulation deviation parameters; drive the crop growth-fertilizer response model based on the fertilizer regulation deviation parameters to re-optimize the fertilizer scheme, so as to output the optimal fertilizer regulation scheme corresponding to each growth stage and synchronize it to the fertilization execution system, thereby realizing the dynamic visualization and execution of agricultural crop fertilizer regulation.

[0006] The beneficial effects of this invention are: The intelligent fertilizer regulation method for agricultural crop growth stages proposed in this invention, based on multiple factors, offers several advantages over existing technologies. Firstly, it simultaneously acquires historical data on crop growth, soil environment, meteorology, and fertilizer application, covering all influencing factors of fertilizer regulation. Growth status data reflects the crop's current nutrient requirements, soil environment data demonstrates nutrient supply capacity, meteorological data reveals the impact of rainfall and temperature on nutrient absorption, and historical fertilizer data provides a reference for scheme optimization. Secondly, through the fusion and standardization of multi-source heterogeneous data, a structured dataset is generated. This multi-source data integration and preprocessing not only compensates for the data fragmentation deficiencies of existing methods but also ensures data consistency and accuracy. This allows subsequent regulation scheme formulation to no longer rely on a single indicator but rather on a multi-dimensional correlation between "crop-soil-meteorology-fertilizer," thus avoiding fertilization deviations caused by incomplete data from the outset. Secondly, by extracting the patterns of physiological indicator changes at different stages from crop growth data, the differences in nutrient requirements at each stage are clarified; by combining soil data to mine the correlation between fertility and nutrient absorption efficiency, the blind application of fertilizers is avoided; a four-dimensional regulation matrix is ​​constructed by introducing meteorological and fertilizer data to quantify the complex relationship between "crop growth - soil fertility - meteorological influence - fertilizer application". For example, under meteorological conditions with high rainfall, the timing of nitrogen fertilizer application needs to be adjusted to reduce loss. This matrix model of multi-factor collaborative analysis solves the problem of one-sided regulation logic in existing methods and provides scientific support for subsequent model construction and scheme formulation. Then, by constructing a response model based on a four-dimensional regulation matrix, the crop growth process under different fertilization schemes can be simulated, generating a simulation dataset. Through coupled analysis of simulation data and actual growth data, the initial fertilization scheme set for each growth stage can be determined. For example, in the seedling stage, a low-nitrogen, high-phosphorus scheme can be selected based on simulation results to promote root development while avoiding excessive phosphorus imbalance in the soil. This "model simulation + data coupling" design allows the fertilization scheme to be fully validated before implementation, avoiding the rigidity of fixed formulas and predicting the impact of the scheme on crop growth and soil fertility in advance. This ensures that the scheme meets the crop's nutrient requirements while preventing soil nutrient imbalance, solving the problem of poor adaptability of existing methods. Finally, by designing a dynamic adjustment mechanism, real-time crop growth data is compared with simulation data to generate control deviation parameters. Based on these deviation parameters, the response model is re-optimized, outputting the optimal solution for each stage and synchronizing it to the fertilization execution system. This closed-loop mechanism allows the fertilization plan to adapt to crop growth changes and environmental fluctuations in real time, avoiding the limitations of existing methods that are "unchanging." It ensures that fertilizer regulation is always precisely matched with the actual needs of the crop throughout the entire growth period, thereby maximizing fertilizer utilization efficiency and crop yield. Attached Figure Description

[0007] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the steps of the intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to the present invention. Figure 2 for Figure 1 A detailed flowchart of step S1. Detailed Implementation

[0008] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0009] To achieve the above objectives, please refer to Figures 1 to 2 This invention provides a method for intelligent fertilizer regulation during the growth stages of agricultural crops based on multiple factors. In the embodiments of this invention, please refer to... Figure 1 The diagram shown is a flowchart illustrating the steps of the intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to the present invention. In this example, the intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors includes the following steps: Step S1: Obtain the growth status data, soil environment data, meteorological data and fertilizer application history data of agricultural crops, and perform multi-source heterogeneous data fusion and standardization processing to generate standardized crop growth characteristic datasets, soil environment characteristic datasets, meteorological impact datasets and fertilizer application characteristic datasets. In this embodiment of the invention, agricultural crop-related data is acquired using multi-source data acquisition tools. Growth status data is collected using growth monitoring equipment, including plant height (10-120cm, measured every 5 days), stem diameter (2-15mm), number of leaves (2-30), and chlorophyll SPAD value (30-60), collected for a total of 120 days. Soil environmental data is collected using soil testing equipment, including nitrogen (80-180mg / kg), phosphorus (20-60mg / kg), potassium (100-200mg / kg) content, pH value (4.0-8.5), and moisture content (12...). Data was collected every 10 days for a total of 12 times, with a range of -30%. Meteorological data was obtained through a combination of meteorological stations and satellite remote sensing, including average daily sunshine duration (3-10h), cumulative precipitation (10-50mm / week), diurnal temperature range (5-15℃), and frequency of extreme weather events (0-2 times / month), recorded daily. Historical fertilizer application data was extracted from the agricultural production management system, including the application dosage, time (11th day of seedling stage, 31st day of jointing stage), and method (broadcasting, drip irrigation) of urea (15-25kg / mu), superphosphate (10-15kg / mu), and potassium chloride (8-12kg / mu). Heterogeneous data were integrated using data fusion tools and aligned by time nodes. Data was processed using standardization tools, with growth status data compressed to 0-1 by (actual value - minimum value) / (maximum value - minimum value), and soil and meteorological data were processed similarly. Finally, four standardized feature datasets were generated, each containing 120 time series data points.

[0010] Step S2: Extract the physiological index change patterns of crops at different growth stages based on crop growth characteristic datasets, and explore the correlation between soil fertility and crop nutrient absorption efficiency by combining soil environmental characteristic datasets. At the same time, introduce meteorological impact datasets and fertilizer application characteristic datasets to construct a four-dimensional regulation matrix for agricultural crops. In this embodiment of the invention, a physiological pattern extraction tool is used to divide the crop growth characteristic dataset into four stages: sowing period (1-10 days), seedling stage (11-30 days), jointing stage (31-60 days), flowering stage (61-90 days), and fruiting stage (91-120 days). The rate of change of physiological indicators at each stage is calculated. For example, if the seedling height increases from 15cm to 25cm, the rate of change is approximately 33.3% (25-15) / 30×100%), thus clarifying the pattern of indicator changes at each stage. A correlation mining tool is used to analyze the relationship between soil environmental characteristic data and crop nutrient absorption efficiency. For example, the absorption efficiency is 34.1% when the nitrogen content is 130mg / kg and 38.5% when it is 150mg / kg, showing a positive correlation between nitrogen content and absorption efficiency (correlation coefficient 0.82). We introduced meteorological impact datasets (e.g., high photosynthetic efficiency with 7 hours of sunlight per day) and fertilizer application characteristic datasets (e.g., 68% utilization efficiency with 20 kg / mu of urea). Using a matrix construction tool, we constructed a 5×3×3×12 four-dimensional regulation matrix for agricultural crops, with growth stages (5) as rows, soil fertility (nitrogen / phosphorus / potassium divided into 3 levels) as columns, meteorological conditions (sunlight / precipitation / temperature difference divided into 3 levels) as depth, and fertilizer schemes (4 types × 3 dosages = 12 types) as the fourth dimension. We filled in the parameters and correlations of each dimension. The matrix elements are the predicted crop growth values ​​(e.g., plant height growth rate) under each combination.

[0011] Step S3: Construct a crop growth-fertilizer response model based on the four-dimensional regulation matrix of agricultural crops, and simulate the crop growth process under different fertilizer application schemes based on the crop growth-fertilizer response model to generate a fertilizer regulation simulation dataset; perform regulation coupling analysis based on the fertilizer regulation simulation dataset and the crop growth characteristic dataset to determine the initial fertilizer application scheme set corresponding to each growth stage that meets the crop nutrient requirements and avoids soil nutrient imbalance. In this embodiment of the invention, a crop growth-fertilizer response model is constructed using a model building tool based on a four-dimensional regulation matrix for agricultural crops. The model includes four modules: crop growth simulation (simulating plant height / chlorophyll changes), soil nutrient migration (simulating nitrogen / phosphorus / potassium transformation and loss), meteorological impact simulation (simulating the effects of light / precipitation), and fertilizer effect calculation (quantifying supply efficiency). Data is shared between the modules. Eighteen orthogonal fertilizer schemes (3 ratios × 3 dosages × 2 intervals) are designed and input into the model for simulation. The simulation period is set to 120 days. The daily outputs include crop growth curves (plant height reaches 65cm in 60 days), soil nutrient change curves (nitrogen decreases from 130mg / kg to 95mg / kg in 120 days), and nutrient absorption curves (total nitrogen absorption of 12kg / acre). Curve features (peak growth rate of 1.2cm / day, nutrient maintenance period of 40 days) are extracted to generate a fertilizer regulation simulation dataset. Using a regulation coupling analysis tool, the demand thresholds (e.g., nitrogen demand of 8-10 kg / mu at the jointing stage) of the simulation dataset are compared with those of the crop growth characteristic dataset. Schemes that meet the demand and have soil nutrient balance within the threshold (90-150 mg / kg) are selected, such as scheme 8 (urea 18 kg / mu + superphosphate 12 kg / mu + potassium chloride 8 kg / mu). The initial fertilizer application scheme set corresponding to the five growth stages is determined (a total of five schemes, one for each stage).

[0012] Step S4: Design a dynamic adjustment mechanism based on the initial fertilizer application scheme set to compare and analyze the real-time crop growth status data with the fertilizer regulation simulation dataset to generate fertilizer regulation deviation parameters; drive the crop growth-fertilizer response model based on the fertilizer regulation deviation parameters to re-optimize the fertilizer scheme, so as to output the optimal fertilizer regulation scheme corresponding to each growth stage and synchronize it to the fertilization execution system, thereby realizing the dynamic visualization and execution of agricultural crop fertilizer regulation.

[0013] In this embodiment of the invention, a dynamic adjustment mechanism is designed using a mechanism design tool based on an initial fertilizer application scheme set. This mechanism includes three levels of control (deviations of 0-5 / 6-10 / >10 corresponding to adjustments of ±5% / ±10% / ±15%), stage deviation triggering conditions (e.g., plant height deviation of ±2cm at the jointing stage), and monthly model recalibration rules. Data from day 40 of the jointing stage (plant height 63cm, chlorophyll 42, soil nitrogen 115mg / kg) is obtained through real-time monitoring equipment and compared with simulated data (plant height 67cm, chlorophyll 48, soil nitrogen 135mg / kg). The deviation parameter calculated using a deviation calculation tool is 9.6, exceeding the acceptable range (0-5), matching a level 2 deviation. The crop growth-fertilizer response model is then re-optimized, adjusting the urea supply efficiency to 72% and the nitrogen leaching coefficient to 0.04. A new scheme is then generated through simulation (urea 19.8kg / mu, superphosphate 13.2kg / mu, potassium chloride 8.8kg / mu). The scheme is converted into control commands (fertilizer spreading speed 2m / s, row spacing 5m) using an instruction conversion tool and pushed to the intelligent fertilizer spreader; the comparison before and after adjustment (plant height prediction 67cm→64cm) is displayed through a visualization platform, and a dynamic report is generated to realize the dynamic visualization and execution of fertilizer regulation.

[0014] Furthermore, as an embodiment of the present invention, reference is made to... Figure 2 As shown, Figure 1 A detailed flowchart of step S1 is shown below. In this embodiment, step S1 includes the following steps: Step S11: Collect growth status data of agricultural crops, including plant height, stem diameter, number of leaves, leaf area, and chlorophyll SPAD value, through crop growth monitoring equipment. In this embodiment of the invention, a portable crop growth monitor is used to set up monitoring points in a 5m × 5m grid in the crop planting area. Three uniformly growing crops are selected as samples at each monitoring point. The monitor's built-in laser ranging module is used to align with the crop from the base to the top growth point to record plant height data; the monitor's diameter measuring probe is placed against the middle of the crop stem to record stem diameter data; the number of leaves is counted manually for each plant to ensure that no new leaflets or aging leaves are missed; the monitor's image scanning function is used to photograph the front of the leaves, and the built-in image analysis module automatically calculates the leaf outline area to generate leaf area data; the monitor's chlorophyll detection probe is placed against the veinless area in the middle of the leaf, held for 2 seconds, and the chlorophyll SPAD value is read. Each leaf is tested three times and the average value is taken to obtain the plant height, stem diameter, number of leaves, leaf area, and chlorophyll SPAD value growth status data for each crop.

[0015] Step S12: Collect soil environmental data including soil nitrogen, phosphorus and potassium content, soil pH value, soil moisture content and soil bulk density using soil testing equipment; In this embodiment of the invention, soil samples were collected in the planting area using a soil sampler in a 10m × 10m grid. Each sampling point employed a stratified sampling method, collecting soil samples at depths of 0-20cm, 20-40cm, and 40-60cm, with 200 grams collected at each depth. The samples were then mixed and placed into soil sample bags. The soil samples were allowed to air dry to constant weight, and after removing stones, roots, and other impurities, they were crushed and passed through a 2mm soil sieve. A soil nitrogen, phosphorus, and potassium analyzer was used. A 5-gram sieved soil sample was placed in the instrument's detection chamber, along with the instrument's reagents. After shaking for 10 minutes and letting it stand for 5 minutes, the nitrogen, phosphorus, and potassium content in the soil was detected using the instrument's spectral analysis function. A soil pH meter was used; the probe was inserted into a soil suspension prepared by mixing another 5 grams of sieved soil with 10 ml of distilled water. After standing for 3 minutes, the pH value was read. A soil moisture meter was used; a 10-gram sieved soil sample was placed in the instrument's heating chamber and heated at 105℃ for 4 hours. The instrument automatically weighed the sample to calculate the soil moisture evaporation and thus the soil moisture content. A soil bulk density meter was used; an undisturbed soil sample was selected, and the sample volume and weight were measured to calculate the soil bulk density. Finally, the soil nitrogen, phosphorus, and potassium content, soil pH value, soil moisture content, and soil bulk density data for each sampling point were obtained.

[0016] Step S13: Collect meteorological data corresponding to the average daily sunshine duration, cumulative precipitation, diurnal temperature range, and frequency of extreme weather events in agricultural crop planting areas by combining meteorological stations and satellite remote sensing. In this embodiment of the invention, a small weather station is deployed at the center of the planting area. The station is equipped with a light sensor, a rain gauge, and a temperature sensor. The light sensor records the light intensity hourly, and the duration of light intensity exceeding 1000 lux over 24 hours is taken as the average daily sunshine duration. The rain gauge automatically records rainfall after each rainfall event, summarizing the daily rainfall at 24:00 and accumulating the cumulative precipitation. The temperature sensor records the air temperature hourly, calculating the difference between the daily highest and lowest temperatures as the diurnal temperature range. Simultaneously, satellite remote sensing images covering the planting area are acquired every three days. Spectral analysis of the remote sensing images identifies extreme weather phenomena such as heavy rain, strong winds, and hail within the planting area. Combined with extreme weather time data recorded by the weather station, the monthly frequency of extreme weather events is calculated. Finally, the weather station and satellite remote sensing data are integrated to generate meteorological data corresponding to the average daily sunshine duration, cumulative precipitation, diurnal temperature range, and frequency of extreme weather events in the planting area.

[0017] Step S14: Extract fertilizer application records from the agricultural production management system for the historical planting cycle of agricultural crops, including fertilizer type, dosage for each application, application time, and application method corresponding to historical fertilizer application data. In this embodiment of the invention, by accessing the database backend of the agricultural production management system, fertilizer application records for the past three planting cycles of the planting area are retrieved. The records are categorized by planting cycle, and specific information on each fertilizer application within each cycle is extracted to determine the fertilizer type as organic fertilizer, nitrogen fertilizer, phosphorus fertilizer, potassium fertilizer, or compound fertilizer. The fertilizer application amount for each application is recorded, and the dosage is determined in kilograms. Based on the agricultural operation timestamps within the system, the specific year, month, and day of each fertilizer application are determined. Through the agricultural operation notes within the system, the application method is clarified as broadcasting, strip application, hole application, or drip irrigation. Finally, historical fertilizer application data is compiled, including the fertilizer type, dosage, application time, and application method for each planting cycle.

[0018] Step S15: Filter outliers and complete the time series data for growth status data, perform spatial interpolation and standardization on soil environmental data, align the time series data and quantify the degree of influence on meteorological data, and encode the features of historical fertilizer application data to generate standardized crop growth characteristic datasets, soil environmental characteristic datasets, meteorological influence datasets, and fertilizer application characteristic datasets.

[0019] In this embodiment of the invention, outliers are identified using box plots for crop growth data. Data exceeding 1.5 times the interquartile range are marked as outliers and removed. For data gaps after outlier removal, linear interpolation is used to fill in the gaps based on data from three consecutive time points before and after the gap, generating a standardized crop growth characteristic dataset. For soil environmental data, Kriging interpolation is used to construct a spatial distribution model of soil environmental data based on data from each sampling point and combined with topographic data of the planting area. Discrete sampling point data are interpolated into continuous areal data, and then Min-Max normalization is used to compress the data to the 0-1 interval, generating a standardized soil environmental characteristic dataset. For meteorological data, the daily time unit of the crop growth cycle is used to uniformly process meteorological data from different collection frequencies. The system is designed to scale up to the daily level. For example, hourly sunshine data is aggregated into daily sunshine duration. Based on the sensitivity of crops to meteorological factors at different growth stages, different weight coefficients are assigned to the average daily sunshine duration, cumulative precipitation, diurnal temperature range, and frequency of extreme weather events. The influence of each meteorological factor is quantified through weighted calculation, generating a meteorological impact dataset. For historical fertilizer application data, fertilizer type is converted into binary code, such as 0001 for organic fertilizer, 0010 for nitrogen fertilizer, 0100 for phosphate fertilizer, 1000 for potassium fertilizer, and 1111 for compound fertilizer. The application time is converted into the number of days since the start of planting, and the application method is converted into numerical code, such as 1 for broadcasting, 2 for strip application, 3 for hole application, and 4 for drip irrigation. The original application dosage is retained, generating a standardized fertilizer application feature dataset.

[0020] Furthermore, step S2 includes the following steps: Step S21: Extract time-series data of physiological indicators corresponding to each growth stage of the crop, including sowing period, seedling stage, jointing stage, flowering stage and fruiting stage, from the crop growth characteristic dataset, and calculate the growth rate and coefficient of variation of the physiological indicators of each stage to generate a crop growth stage feature vector. In this embodiment of the invention, a stage data extraction tool is used to filter the time-series data of physiological indicators for the sowing period (1-10 days), seedling stage (11-30 days), jointing stage (31-60 days), flowering stage (61-90 days), and fruiting stage (91-120 days) within the complete crop growth cycle from the crop growth characteristic dataset. Each stage is divided into nodes of 5 days, and data for each stage are extracted at nodes 2, 4, 6, 6, and 6 respectively. The indicators include plant height (unit: cm), stem diameter (unit: mm), number of leaves (unit: leaf), and chlorophyll SPAD value. A growth rate calculation tool is used to calculate the growth rate of each stage according to the formula (indicator value of the next node - indicator value of the previous node) / time interval (5 days). For example, if the plant height increases from 15cm to 25cm during the seedling stage, the growth rate = (25-15) / 5 = 2cm / day. Using a coefficient of variation (CV) calculation tool, the CV for each stage was calculated using the formula (standard deviation of the index / average value of the index) × 100%. For example, the average number of leaves at the jointing stage was 8, with a standard deviation of 1.2, and the CV = (1.2 / 8) × 100% = 15%. The growth rate of each stage (5 stages × 4 indicators = 20 data points) and the CV (20 data points) were arranged in stage order to generate a 40-dimensional crop growth stage feature vector.

[0021] Step S22: Based on the soil environmental feature dataset, analyze the correlation between soil nitrogen, phosphorus and potassium content and crop chlorophyll content and leaf nitrogen content, and generate soil fertility-nutrient absorption correlation parameters by calculating the contribution coefficient of soil nutrients to crop nutrient absorption; at the same time, calculate the mapping relationship between soil pH value and soil nutrient availability and combine it with the soil fertility-nutrient absorption correlation parameters to generate soil fertility-nutrient feature vector. In this embodiment of the invention, correlation analysis tools were used to calculate the correlation between nitrogen (80-180 mg / kg), phosphorus (20-60 mg / kg), and potassium (100-200 mg / kg) content data in the soil environmental characteristic dataset and chlorophyll SPAD values ​​(30-60) and leaf nitrogen content (2.0-4.5%) in the crop growth characteristic dataset. The correlation coefficients were found to be 0.82 for nitrogen and 0.88 for nitrogen and leaf nitrogen content, 0.65 for phosphorus and 0.52 for phosphorus and leaf nitrogen content, 0.71 for potassium and 0.61 for potassium and leaf nitrogen content. The contribution coefficient calculation tool was used to calculate the contribution coefficient according to the formula (correlation coefficient × average soil nutrient value) / average crop index value. For example, the nitrogen contribution coefficient = (0.88 × 130 mg / kg) / 3.25% = 35.2, the phosphorus contribution coefficient = (0.52 × 40 mg / kg) / 3.25% = 6.4, and the potassium contribution coefficient = (0.61 × 150 mg / kg) / 3.25% = 28.4. These three were used as soil fertility-nutrient absorption correlation parameters. Combined with the previously constructed pH-nutrient availability mapping table, such as nitrogen availability of 0.8, phosphorus of 0.7, and potassium of 0.8 at pH 6.0, the available nutrient content (nitrogen 130 × 0.8 = 104 mg / kg, etc.) was calculated. This was then combined with the soil fertility-nutrient absorption correlation parameters to generate a 6-dimensional soil fertility-nutrient feature vector (nitrogen, phosphorus, potassium contribution coefficients and available content).

[0022] Step S23: Calculate the correlation between sunshine duration, temperature, precipitation and crop growth rate in the meteorological impact dataset to obtain the influence weight coefficient of each meteorological factor on crop growth; Based on the influence weight coefficient of each meteorological factor on crop growth and combined with the water and light requirements of crops at different growth stages, generate meteorological factor-growth response correlation parameters. In this embodiment of the invention, a correlation analysis was performed using a correlation calculation tool to correlate the daily average sunshine duration (3-10h), diurnal temperature range (5-15℃), and cumulative precipitation (10-50mm / week) in the meteorological impact dataset with the growth rate (e.g., plant height 2cm / day, stem diameter 0.3mm / day) in the crop growth stage feature vector. The correlation coefficients were found to be 0.75 between sunshine duration and plant height growth rate, 0.68 between temperature range and stem diameter growth rate, and 0.62 between precipitation and leaf growth rate. Using a weighting coefficient calculation tool, the influence weighting coefficients were calculated using the formula (correlation coefficient / sum of the three correlation coefficients) × 100%. For example, the sunshine weight = (0.75 / (0.75+0.68+0.62)) × 100% = 36.6%, the temperature range weight = 33.0%, and the precipitation weight = 30.4%. Combining the water and light requirements of crops (seedling stage requires 5-6 hours of light / day and 15-20 mm of water / week; flowering stage requires 8-10 hours of light / day and 30-40 mm of water / week), the weights are adjusted according to the formula (weight coefficient × stage requirement / average requirement). For example, the light weight during the flowering stage = 36.6% × (9h / 7h) = 47.3%. Finally, the weight coefficients of the three meteorological factors at each stage are generated, forming a 15-dimensional (5 stages × 3 factors) meteorological factor-growth response correlation parameter.

[0023] Step S24: Extract crop yield and quality indicators corresponding to different fertilizer types and application doses from the fertilizer application feature dataset, calculate fertilizer utilization efficiency, and generate fertilizer effect parameters. In this embodiment of the invention, four fertilizer types—urea (46% nitrogen), superphosphate (16% phosphorus), potassium chloride (60% potassium), and compound fertilizer (15% nitrogen, 15% phosphorus, and 15% potassium)—are extracted from a fertilizer application characteristic dataset using a data filtering tool. Three application doses are set for each type (e.g., urea 15, 20, and 25 kg / mu), corresponding to the extracted crop yield (kg / mu) and quality indicators (soluble sugar content, %). For example, with urea at 20 kg / mu, the yield is 650 kg / mu, and the soluble sugar content is 8.2%. A fertilizer utilization efficiency calculation tool is used to calculate the efficiency using the formula (yield after fertilization - yield of blank control) / fertilizer application rate. For example, if the blank control yield is 500 kg / mu, the efficiency of urea at 20 kg / mu is (650-500) / 20 = 7.5 kg / kg. The yield (3 data points), soluble sugar content (3 data points), and utilization efficiency (3 data points) corresponding to the 3 dosages of each fertilizer type were arranged according to fertilizer type to generate 36-dimensional fertilizer effect parameters (4 fertilizers × 3 indicators × 3 dosages).

[0024] Step S25: Perform dimensional matching on the crop growth stage feature vector, soil fertility-nutrient feature vector, meteorological factor-growth response correlation parameters, and fertilizer effect parameters to construct a four-dimensional regulation matrix for agricultural crops. Each element in this matrix represents the contribution of the combination of soil, meteorological conditions, and fertilizer schemes to the regulation of crop growth at a certain growth stage.

[0025] In this embodiment of the invention, a dimension matching tool is used to unify the dimensions of crop growth stage feature vectors (40 dimensions), soil fertility-nutrient feature vectors (6 dimensions), meteorological factors-growth response correlation parameters (15 dimensions), and fertilizer effect parameters (36 dimensions). Zero-padding is then used to supplement the dimensions of each vector to a total of 40 dimensions, ensuring consistency in the total dimensions. Using a matrix construction tool, a 5×3×3×12 four-dimensional agricultural crop regulation matrix is ​​constructed, with crop growth stages (5 stages) as rows, soil fertility levels (divided into 3 levels based on nitrogen, phosphorus, and potassium content) as columns, meteorological conditions (divided into 3 levels based on light / temperature difference / precipitation) as depth, and fertilizer schemes (4 types × 3 dosages = 12 types) as the fourth dimension. The contribution of each element in the matrix is ​​calculated using the formula (value of the feature vector corresponding to the growth stage × value of the feature vector corresponding to the soil × value of the meteorological parameter × value of the fertilizer parameter) / the total average product. For example, the contribution of seedling stage (growth vector value 0.2), high fertility (soil vector value 0.8), suitable weather (meteorological parameter value 0.9), and urea 20kg / mu (fertilizer parameter value 0.7) is (0.2 × 0.8 × 0.9 × 0.7) / 0.1008 = 1.0, which finally forms a four-dimensional regulation matrix containing 540 elements.

[0026] Furthermore, step S22 includes the following steps: Step S221: Extract time-series data of soil nitrogen, phosphorus and potassium content for multiple consecutive growth cycles from the soil environmental characteristics dataset, and extract nutrient absorption-related indicators for the corresponding cycles, including crop chlorophyll SPAD value, leaf nitrogen content and crop biomass, from the crop growth characteristics dataset. In this embodiment of the invention, a data filtering tool is used to select soil nitrogen, phosphorus, and potassium content data for five complete planting cycles from a standardized soil environmental characteristic dataset. Time nodes are then divided according to the growth stages of each planting cycle (e.g., seedling stage, jointing stage, grain-filling stage, maturity stage), and soil nitrogen, phosphorus, and potassium content data corresponding to each time node are extracted, forming a time-series data of soil nitrogen, phosphorus, and potassium content for multiple consecutive growth cycles. Each cycle contains 12 time nodes (2 nodes per month). Simultaneously, a data matching tool is used to extract crop data from the crop growth characteristic dataset corresponding to the same time nodes as the soil nitrogen, phosphorus, and potassium content time-series data within the five planting cycles. Chlorophyll SPAD values ​​are directly extracted from the corresponding detection data, leaf nitrogen content is obtained through chemical analysis of leaf samples (three crop leaves are sampled and analyzed at each time node), and crop biomass is obtained through the fresh weight of the aboveground parts of the crop (five crops are weighed at each time node, and the average value is taken). Finally, a dataset of nutrient absorption-related indicators for the corresponding cycle is formed.

[0027] Step S222: Using time-series data of soil nitrogen, phosphorus and potassium content as independent variables and nutrient absorption-related indicators as dependent variables, construct a multiple linear regression model and calculate the model regression coefficients. The model regression coefficients are defined as the contribution coefficients of soil fertility to nutrient absorption. In this embodiment of the invention, soil nitrogen, phosphorus, and potassium content time-series data for five consecutive planting cycles are used as independent variables using data processing tools. Independent variables X1 represent soil nitrogen content, X2 represent soil phosphorus content, and X3 represent soil potassium content, with each independent variable containing 60 data points (5 cycles × 12 nodes). Nutrient absorption-related indicators for the corresponding cycles are used as dependent variables, with dependent variables Y1 representing chlorophyll SPAD value, Y2 representing leaf nitrogen content, and Y3 representing crop biomass, each also containing 60 data points. Multiple linear regression analysis tools are used to construct regression models for Y1 with X1, X2, and X3, Y2 with X1, X2, and X3, and Y3 with X1, X2, and X3, respectively. The regression coefficients for each model are calculated using the tool; for example, Y1 = a1X1 + a2X2 + a3X3 + b1 (where a1, a2, and a3 are regression coefficients, and b1 is a constant term). Similarly, the regression coefficients for models Y2 and Y3 are obtained. The regression coefficients corresponding to the independent variables in all models (corresponding coefficients of models a1, a2, a3 and Y2, Y3) are summarized and defined as the contribution coefficients of soil fertility-nutrient absorption basis. The coefficients corresponding to nitrogen content are set in the range of 0.3-0.8, the coefficients corresponding to phosphorus content are set in the range of 0.1-0.4, and the coefficients corresponding to potassium content are set in the range of 0.2-0.5.

[0028] Step S223: Extract soil moisture content from the soil environmental characteristic dataset and analyze the impact of soil moisture content on soil nutrient dissolution and crop root absorption. When the soil moisture content is too low, calculate the decrease in nutrient dissolution rate; when the soil moisture content is too high, calculate the nutrient leaching risk coefficient. At the same time, use the decrease in nutrient dissolution rate and the nutrient leaching risk coefficient as correction terms to correct the basic contribution coefficient of soil fertility-nutrient absorption, and obtain the soil fertility-nutrient absorption correlation parameters. In this embodiment of the invention, soil moisture content data for each time point within five planting cycles is extracted from a soil environmental characteristic dataset using a data extraction tool. Each cycle contains 12 nodes, totaling 60 moisture content data points. Data analysis tools determine the suitable soil moisture content range for crops to be 18%-25%. When the soil moisture content is below 18%, the decrease in nutrient solubility is calculated as (suitable moisture content - actual moisture content) / suitable moisture content × solubility decrease coefficient. When the soil moisture content is above 25%, a nutrient leaching risk coefficient is calculated as (actual moisture content - suitable moisture content) / suitable moisture content × leaching risk increase coefficient. The decrease in nutrient solubility is then converted into a correction coefficient (1 - decrease / 100), e.g., a decrease of 0.167 corresponds to a correction coefficient of 0.833. The nutrient leaching risk coefficient is directly used as a correction term (correction coefficient is 1 - risk coefficient), e.g., a risk coefficient of 0.096 corresponds to a correction coefficient of 0.904. Using the coefficient correction tool, the basic contribution coefficient of soil fertility-nutrient absorption is multiplied by the correction coefficient at the corresponding time point to obtain the corrected soil fertility-nutrient absorption correlation parameter. For example, when the basic contribution coefficient a1 is 0.5 and the correction coefficient is 0.833, the correlation parameter is 0.5×0.833=0.4165.

[0029] Step S224: Extract soil pH values ​​from the soil environmental characteristic dataset, and establish a mapping relationship table between soil pH values ​​and the availability of different nutrients by combining crop nutrient absorption characteristics; In this embodiment of the invention, soil pH data from the past five planting cycles were extracted using a data acquisition tool, totaling 80 data points (16 detection points per cycle), with a statistical pH range of 4.5-8.5. Combining crop nutrient absorption characteristics (e.g., optimal absorption of nitrogen, phosphorus, and potassium at pH 6.0-7.0), a mapping table between soil pH and the availability of different nutrients was established using a data association analysis tool. The mapping table divides pH values ​​into eight intervals of 0.5 (4.5-5.0, 5.0-5.5, 5.5-6.0, 6.0-6.5, 6.5-7.0, 7.0-7.5, 7.5-8.0, 8.0-8.5), with each interval corresponding to the availability level of nitrogen, phosphorus, and potassium. For example, in the pH range of 4.5-5.0, nitrogen availability is 0.6, phosphorus availability is 0.4, and potassium availability is 0.7; in the pH range of 6.0-6.5, nitrogen availability is 1.0, phosphorus availability is 0.9, and potassium availability is 0.9; and in the pH range of 8.0-8.5, nitrogen availability is 0.7, phosphorus availability is 0.5, and potassium availability is 0.8. All availability values ​​are determined based on the ratio of the actual absorption by the crop to the optimal absorption in the corresponding pH range.

[0030] Step S225: Calculate the nutrient availability coefficient under different pH ranges based on the mapping relationship table, and multiply the nutrient availability coefficient by the soil nutrient content to obtain the soil available nutrient content; generate soil acid-base adjustment parameters based on the deviation between the soil available nutrient content and the actual absorption by crops; and generate soil fertility-nutrient feature vector by combining the soil fertility-nutrient absorption correlation parameters.

[0031] In this embodiment of the invention, a mapping query tool is used to look up the nutrient availability coefficients for the corresponding pH range in the mapping table based on the soil pH value at each time point in the soil environmental characteristic dataset. For example, if the pH value at a certain time point is 5.8, corresponding to the pH range of 5.5-6.0, the nitrogen availability coefficient is 0.8, the phosphorus availability coefficient is 0.7, and the potassium availability coefficient is 0.8. A data calculation tool is then used to multiply the soil nitrogen, phosphorus, and potassium contents at that time point by their corresponding availability coefficients to obtain the available nutrient content of the soil. For example, if the soil nitrogen content is 150 mg / kg, the available nitrogen content is 150 × 0.8 = 120 mg / kg. Similarly, the available phosphorus and available potassium contents are calculated. Obtain the standard nutrient uptake of crops at the corresponding growth stages from the crop nutrient requirement database. For example, the standard nitrogen uptake of seedling crops is 100 mg / kg. Calculate the deviation between the available soil nutrient content and the standard amount. For example, when the available nitrogen content is 120 mg / kg, the deviation is +20 mg / kg. Based on the magnitude of the deviation, generate soil pH adjustment parameters. When the absolute value of the deviation is greater than 30 mg / kg, the adjustment parameter is 0.2 (requires emphasis); when the absolute value of the deviation is 10-30 mg / kg, the adjustment parameter is 0.1 (requires weak adjustment); when the absolute value of the deviation is less than 10 mg / kg, the adjustment parameter is 0 (no adjustment required). Using a vector synthesis tool, soil pH adjustment parameters are combined with soil fertility-nutrient absorption correlation parameters (nitrogen, phosphorus, and potassium correlation parameters) at the corresponding time points to form a soil fertility-nutrient feature vector containing four elements. For example, if the adjustment parameter is 0.1, the nitrogen correlation parameter is 0.4165, the phosphorus correlation parameter is 0.22, and the potassium correlation parameter is 0.35, the feature vector is [0.4165, 0.22, 0.35, 0.1].

[0032] Furthermore, step S223 includes the following steps: Real-time monitoring data of soil moisture content is extracted from the soil environmental characteristics dataset, and combined with the suitable moisture content range of the crop growth stage, the deviation value between the actual moisture content and the suitable moisture content is calculated to obtain the moisture content deviation parameter. In this embodiment of the invention, real-time data on soil moisture content at the current growth stage of the crop (taking the jointing stage as an example) is obtained from a soil environmental characteristic dataset using a real-time data extraction tool. The data collection frequency is set to once every 2 hours, and 12 real-time data points are obtained after 24 hours of continuous collection, such as 18.5%, 17.8%, and 19.2%. The suitable soil moisture content range for the crop at the jointing stage is determined to be 19%-24% using a crop growth stage database. The deviation value between each real-time moisture content data point and the suitable moisture content range is calculated using a data calculation tool. The deviation value is calculated by subtracting the median suitable moisture content from the actual moisture content ((19%+24%) / 2=21.5%). For example, when the actual moisture content is 18.5%, the deviation value is 18.5%-21.5%=-3%; when the actual moisture content is 25%, the deviation value is 25%-21.5%=3.5%. The deviation values ​​corresponding to all real-time data are summarized to obtain a moisture content deviation parameter dataset.

[0033] Furthermore, a nutrient solubility model is constructed based on the moisture content deviation parameter. When the real-time soil moisture content is lower than or equal to the suitable moisture content range, the nutrient solubility decreases linearly with decreasing moisture content, and the decrease in nutrient solubility is calculated as (suitable moisture content - actual moisture content) / suitable moisture content × solubility decrease coefficient. When the real-time soil moisture content is higher than the suitable moisture content range, the nutrient leaching risk increases exponentially with increasing moisture content, and the nutrient leaching risk coefficient is calculated as (actual moisture content - suitable moisture content) / suitable moisture content × leaching risk increase coefficient. In this embodiment of the invention, a nutrient solubility model is built using a model building tool based on the moisture content deviation parameter. The solubility reduction coefficient is set to 1.2. When the real-time soil moisture content is lower than or equal to the suitable moisture content range (≤24%), the nutrient solubility decreases linearly with decreasing moisture content. Using the median suitable moisture content of 21.5% as a baseline (where the nutrient solubility is 100%), the decrease in nutrient solubility is calculated. For example, if the actual moisture content is 18.5% (3% lower than the median), the decrease is approximately (21.5% - 18.5%) / 21.5% × 1.2 ≈ 16.7%; if the actual moisture content is 24% (at the upper limit of the suitable range), the decrease is approximately (21.5% - 24%) / 21.5% × 1.2 (since the result is negative, it is taken as 0, i.e., no decrease). The leaching risk escalation coefficient is set at 1.5. When the real-time soil moisture content is higher than the suitable moisture content range (>24%), the nutrient leaching risk increases exponentially with the increase of moisture content. Taking the leaching risk coefficient of 24% as the benchmark, the nutrient leaching risk coefficient is calculated. For example, if the actual moisture content is 26% (2% higher than the suitable upper limit), the risk coefficient is (26%-24%) / 24%×1.5≈0.125; if the actual moisture content is 28% (4% higher than the suitable upper limit), the risk coefficient is (28%-24%) / 24%×1.5≈0.25. Finally, the decrease in nutrient solubility and the nutrient leaching risk coefficient corresponding to each real-time moisture content are obtained.

[0034] Furthermore, the decrease in nutrient solubility rate is weighted and integrated with the nutrient leaching risk coefficient to obtain the comprehensive coefficient of soil moisture impact; In this embodiment of the invention, a weighted fusion tool is used to process the decrease in nutrient solubility and the nutrient leaching risk coefficient. Based on the sensitivity of crops to water at different growth stages, the weight of the decrease in nutrient solubility is set to 0.6 and the weight of the nutrient leaching risk coefficient is set to 0.4. When the real-time moisture content is below the suitable range, only the decrease in nutrient solubility is considered, and the integrated formula is: Soil Moisture Influence Comprehensive Coefficient = 1 - (Decrease in Nutrient Solubility × 0.6). For example, when the decrease is 16.7%, the comprehensive coefficient = 1 - (16.7% × 0.6) ≈ 0.8998. When the real-time moisture content is above the suitable range, only the nutrient leaching risk coefficient is considered, and the integrated formula is: Soil Moisture Influence Comprehensive Coefficient = 1 - (Nutrient Leaching Risk Coefficient × 0.4). For example, when the risk coefficient is 0.125, the comprehensive coefficient = 1 - (0.125 × 0.4) = 0.95. When the real-time moisture content is within the suitable range, the comprehensive coefficient is set to 1.0, and the final soil moisture influence comprehensive coefficient corresponding to each real-time moisture content is obtained.

[0035] Furthermore, the soil fertility-nutrient absorption baseline contribution coefficient is multiplied by the soil moisture influence comprehensive coefficient to obtain the corrected soil fertility-nutrient absorption correlation parameters. In this embodiment of the invention, the basic contribution coefficients of soil fertility-nutrient absorption are extracted from previous calculations. Taking the jointing stage as an example, the basic contribution coefficients for nitrogen content are 0.6, for phosphorus content 0.3, and for potassium content 0.4. Using a coefficient multiplication tool, each basic contribution coefficient is multiplied by the comprehensive coefficient of soil moisture influence corresponding to the real-time moisture content to obtain the corrected soil fertility-nutrient absorption correlation parameters. For example, when the comprehensive coefficient is 0.8998, the nitrogen correlation parameter = 0.6 × 0.8998 ≈ 0.5399, the phosphorus correlation parameter = 0.3 × 0.8998 ≈ 0.2699, and the potassium correlation parameter = 0.4 × 0.8998 ≈ 0.3599; when the comprehensive coefficient is 0.95, the nitrogen correlation parameter = 0.6 × 0.95 = 0.57, the phosphorus correlation parameter = 0.3 × 0.95 = 0.285, and the potassium correlation parameter = 0.4 × 0.95 = 0.38, thus completing the correction of all basic contribution coefficients.

[0036] Furthermore, the correction effect is verified by comparing the differences in the correlation parameters before and after the correction. If the degree of fit between the corrected parameters and the actual nutrient absorption data of the crop exceeds the preset threshold, the corrected soil fertility-nutrient absorption correlation parameters are retained; otherwise, the solubility reduction coefficient and the leaching risk increase coefficient are readjusted until the degree of fit meets the requirements.

[0037] In this embodiment of the invention, a goodness-of-fit verification tool is used to compare the goodness of fit between the soil fertility-nutrient absorption correlation parameters before and after correction and the actual nutrient absorption data of crops (such as leaf nitrogen content and chlorophyll SPAD value). A preset goodness-of-fit improvement threshold of 8% is set. The goodness-of-fit R² between the parameters before correction and the actual data (e.g., 0.72) and the goodness-of-fit R² between the parameters after correction and the actual data (e.g., 0.81) are calculated. The improvement is calculated as (0.81-0.72) / 0.72×100%=12.5%. If this improvement exceeds 8%, the corrected correlation parameters are retained. If the goodness-of-fit improvement does not reach the threshold (e.g., R² after correction is 0.77, improvement is 6.9%), the solubility reduction coefficient (e.g., from 1.2 to 1.3) and the leaching risk increase coefficient (e.g., from 1.5 to 1.6) are readjusted. The above steps are repeated, the comprehensive coefficient and the corrected parameters are recalculated, and the goodness of fit is verified again until the goodness-of-fit improvement exceeds 8%, thus determining the final soil fertility-nutrient absorption correlation parameters.

[0038] Furthermore, the calculation process for the solubility reduction coefficient and the leaching risk increase coefficient includes the following steps: By extracting soil texture type, soil organic matter content and soil porosity data from soil environmental feature datasets, and combining them with the nutrient requirements of crop growth stages, a soil-crop nutrient compatibility assessment matrix is ​​constructed. In this embodiment of the invention, a data classification and extraction tool is used to extract soil texture type (divided into clay, loam, and sandy soil), soil organic matter content (in g / kg, data range 15-35 g / kg), and soil porosity (in %, data range 40%-60%) from the soil environmental feature dataset for the current planting area. For each indicator, data from 30 sampling points are collected and averaged. For example, for clay texture, organic matter content is 22 g / kg, and porosity is 48%. Combined with the nutrient requirements of the crop at different growth stages (taking the grain-filling stage as an example) (high nitrogen requirement, medium phosphorus requirement, medium potassium requirement), a matrix construction tool is used to build a soil-crop nutrient suitability assessment matrix. The matrix uses soil texture type (clay, loam, sandy soil) as the row dimension and nutrient type (nitrogen, phosphorus, potassium) as the column dimension. The matrix elements are the compatibility scores (1-5 points, with 5 points being the best). Clay has a phosphorus compatibility score of 4 points, nitrogen compatibility score of 3 points, and potassium compatibility score of 3 points; loam has a nitrogen compatibility score of 5 points, phosphorus compatibility score of 4 points, and potassium compatibility score of 4 points; and sandy soil has a nitrogen compatibility score of 2 points, phosphorus compatibility score of 2 points, and potassium compatibility score of 3 points. The scores are determined based on the matching degree between the fertilizer retention capacity of different soil textures and the nutrient requirements of crops, ultimately forming a 3×3 soil-crop nutrient compatibility assessment matrix.

[0039] Furthermore, based on the soil-crop nutrient compatibility assessment matrix, the adsorption capacity parameters of different soil textures for nutrients at each growth stage are calculated; the soil nutrient adsorption capacity parameters are obtained by estimating and standardizing the product of soil organic matter content and soil porosity, combined with the corresponding adsorption capacity parameters. In this embodiment of the invention, the adsorption capacity parameters of different soil textures for nutrients at each growth stage are calculated using a parameter calculation tool based on the soil-crop nutrient compatibility assessment matrix. The conversion formula between compatibility score and adsorption capacity parameter is set as follows: Adsorption capacity parameter = (Compatibility score / 5) × 0.8 (0.8 is the basic coefficient of adsorption capacity). Taking the grain-filling stage as an example, the compatibility score of clay for phosphorus is 4 points, and the adsorption capacity parameter = (4 / 5) × 0.8 = 0.64; the compatibility score for nitrogen is 3 points, and the adsorption capacity parameter = (3 / 5) × 0.8 = 0.48; the compatibility score for potassium is 3 points, and the adsorption capacity parameter = (3 / 5) × 0.8 = 0.48. The compatibility score of loam for nitrogen is 5 points, with an adsorption capacity parameter of (5 / 5) × 0.8 = 0.8; the compatibility score for phosphorus is 4 points, with an adsorption capacity parameter of 0.64; and the compatibility score for potassium is 4 points, with an adsorption capacity parameter of 0.64. The compatibility score of sandy soil for nitrogen is 2 points, with an adsorption capacity parameter of (2 / 5) × 0.8 = 0.32; the compatibility score for phosphorus is 2 points, with an adsorption capacity parameter of 0.32; and the compatibility score for potassium is 3 points, with an adsorption capacity parameter of 0.48. Thus, the adsorption capacity parameters for each nutrient corresponding to each soil texture are obtained. Next, the product of soil organic matter content and soil porosity is calculated using data processing tools. Taking clay soil as an example, with an organic matter content of 22 g / kg and a porosity of 48%, the product is 22 × 48 = 1056 (unit: g·% / kg). Multiply this product by the corresponding nutrient adsorption capacity parameter (e.g., 0.64 for phosphorus adsorption capacity in clay), to obtain an estimated value of 1056 × 0.64 = 675.84. Use the Min-Max standardization tool to process the estimated value, setting the standardization range to 0.2-0.8. The standardization formula is: Soil nutrient adsorption capacity parameter = 0.2 + (estimated value - minimum value) / (maximum value - minimum value) × 0.6 (0.6 is the standardization span). Assuming the minimum estimated value for all soil textures and nutrient combinations is 280 and the maximum is 1200, the standardized calculation for phosphorus adsorption in clay is: 0.2 + (675.84 - 280) / (1200 - 280) × 0.6 ≈ 0.2 + 0.395 × 0.6 ≈ 0.437. Similarly, calculate the soil nutrient adsorption capacity parameters for other soil textures and nutrient combinations, ensuring that the parameters are all within the 0.2-0.8 range.

[0040] Furthermore, historical fertilizer types and corresponding nutrient dissolution rate data are extracted from the fertilizer application feature dataset. The dissolution baseline coefficients of different fertilizer types in the current soil texture are calculated, and fertilizer-soil dissolution fit coefficients are generated by combining soil nutrient adsorption capacity parameters. The weighted summation of adsorption capacity parameters and fertilizer-soil dissolution fit coefficients is performed, and the dissolution rate reduction coefficient is generated by inverse exponential correction by introducing the concentration of organic acids secreted by crop roots. In this embodiment of the invention, a data filtering tool is used to extract historically applied fertilizer types (divided into four categories: urea, superphosphate, potassium chloride, and compound fertilizer) and corresponding nutrient dissolution rate data (in % / day; urea nitrogen dissolution rate 8% / day, superphosphate phosphorus dissolution rate 5% / day, potassium chloride potassium dissolution rate 7% / day, and compound fertilizer nitrogen, phosphorus, and potassium dissolution rates 6% / day, 4% / day, and 5% / day, respectively) from the fertilizer application feature dataset. A coefficient calculation tool is used, with a urea nitrogen dissolution rate of 8% / day in the soil as the baseline (dissolution baseline coefficient 1.0), to calculate the dissolution baseline coefficient for different fertilizer types in the current soil texture (e.g., clay). The formula is: Dissolution baseline coefficient = (dissolution rate in the current texture / urea dissolution rate in the soil). The dissolution rate of urea nitrogen in clay is 6% / day, with a dissolution baseline coefficient of 6 / 8 = 0.75; the dissolution rate of superphosphate phosphorus is 4% / day, with a dissolution baseline coefficient of 4 / 8 = 0.5; and the dissolution rate of potassium chloride is 5% / day, with a dissolution baseline coefficient of 5 / 8 = 0.625. Multiplying the dissolution baseline coefficient by the corresponding soil nutrient adsorption capacity parameter (e.g., the adsorption capacity parameter for nitrogen in clay is 0.48) yields the fertilizer-soil solubility compatibility coefficient. For example, the urea-clay nitrogen compatibility coefficient is 0.75 × 0.48 = 0.36, the superphosphate-clay phosphorus compatibility coefficient is 0.5 × 0.64 = 0.32, and the potassium chloride-clay potassium compatibility coefficient is 0.625 × 0.48 = 0.3. Secondly, a weighted summation tool was used, with the adsorption capacity parameter and fertilizer-soil solubility compatibility coefficient each having a weight of 0.5. The two were then weighted and summed. Taking clay phosphorus as an example, the adsorption capacity parameter was 0.64 and the compatibility coefficient was 0.32, so the weighted sum was 0.64 × 0.5 + 0.32 × 0.5 = 0.48. The concentration of organic acids secreted by crop roots during the current growth stage (grain-filling stage) was collected using crop physiological monitoring tools (unit: mmol / L, data range: 2-8 mmol / L, e.g., a detection value of 5 mmol / L). The inverse exponential correction formula was set as: correction factor = e^(-0.1 × organic acid concentration) (0.1 is the correction coefficient). The correction factor was calculated as e^(-0.1 × 5) = e^(-0.5) ≈ 0.6065. Multiply the weighted sum by the correction factor, and add 0.6 (the base value of the solubility reduction coefficient) to obtain the solubility reduction coefficient. For example, the solubility reduction coefficient of clay phosphorus is 0.48 × 0.6065 + 0.6 ≈ 0.291 + 0.6 ≈ 0.891. Similarly, calculate the solubility reduction coefficients for other soil textures and nutrient combinations, ensuring that the coefficients are within the range of 0.6-1.2 to meet the calculation requirements of the subsequent nutrient solubility model.

[0041] Furthermore, soil saturated moisture content, soil permeability coefficient and groundwater depth data were extracted from the soil environmental characteristics dataset, and a soil moisture leakage risk assessment model was constructed by combining historical precipitation intensity and duration data from the meteorological impact dataset. In this embodiment of the invention, a data-oriented extraction tool is used to extract soil saturation moisture content (in %, data range 30%-50%), soil permeability coefficient (in mm / h, data range 0.5-5 mm / h), and groundwater depth (in m, data range 1-5 m) from the soil environmental feature dataset for the current planting area. For each indicator, data from 20 sampling points are collected and averaged, such as a saturation moisture content of 38%, a permeability coefficient of 2.2 mm / h, and a groundwater depth of 2.5 m. Historical precipitation intensity (in mm / h, divided into light rain 0.5-2 mm / h, moderate rain 2-10 mm / h, and heavy rain >10 mm / h) and duration (in hours) data for the past three growth cycles are extracted from the meteorological impact dataset. The average duration corresponding to each precipitation intensity is calculated, such as light rain averaging 6 hours, moderate rain averaging 4 hours, and heavy rain averaging 2 hours. A soil moisture leakage risk assessment model was constructed using a model building tool. The model inputs were soil saturated moisture content, permeability coefficient, groundwater level depth, and rainfall intensity and duration. The output was the leakage risk level (1-5). By setting the weights of each input (saturated moisture content 0.2, permeability coefficient 0.3, groundwater level depth 0.2, rainfall data 0.3), the comprehensive risk value was calculated and mapped to the risk level, ultimately forming a quantifiable soil moisture leakage risk assessment model.

[0042] Furthermore, based on the soil moisture leakage risk assessment model, the soil moisture leakage capacity parameter is estimated by the ratio of soil saturated moisture content to soil permeability coefficient and the leakage buffer coefficient corresponding to the groundwater level depth; historical fertilizer leaching records are extracted from the fertilizer application characteristic dataset, and nutrient leaching sensitivity parameters are calculated based on the ratio of historical leaching amount to corresponding fertilizer application amount and the leaching coefficient of nutrient type. In this embodiment of the invention, soil moisture leakage capacity parameters are calculated using a parameter calculation tool based on a soil moisture leakage risk assessment model. First, the ratio of soil saturated moisture content to soil permeability coefficient is calculated. Taking a saturated moisture content of 38% and a permeability coefficient of 2.2 mm / h as an example, the ratio = 38 ÷ 2.2 ≈ 17.27 (%·h / mm). The leakage buffer coefficient is determined based on the groundwater level depth. The buffer coefficients are set to 0.3 for a groundwater level depth of 1m, 0.5 for 2m, 0.7 for 3m, 0.8 for 4m, and 0.9 for 5m. The current groundwater level depth is 2.5m. The buffer coefficient is calculated using linear interpolation: 0.5 + (2.5 - 2) / (3 - 2) × (0.7 - 0.5) = 0.6. Multiplying the ratio by the buffer coefficient yields the soil moisture permeability parameter: 17.27 × 0.6 ≈ 10.36. The parameter range is set to 5-20. If the calculated result exceeds the range, the boundary value is taken. For example, when the ratio is 25 and the buffer coefficient is 0.8, the parameter = 25 × 0.8 = 20 (taking the upper limit), ensuring that the parameter can be used for subsequent leaching risk calculations. Next, using a data filtering tool, historical fertilizer leaching records for the past three growth cycles are extracted from the fertilizer application characteristic dataset. These records include the leaching amount (in kg / mu) and the corresponding fertilizer application amount (in kg / mu) after each fertilization. For example, if urea is applied at 20 kg / mu, the leaching amount is 1.2 kg / mu; if potassium chloride is applied at 15 kg / mu, the leaching amount is 0.9 kg / mu. The ratio of historical leaching amount to corresponding fertilizer application amount in each record is calculated. For example, the leaching ratio for urea = 1.2 ÷ 20 = 0.06, and the leaching ratio for potassium chloride = 0.9 ÷ 15 = 0.06. Leaching coefficients are set according to nutrient type: nitrogen 1.2, phosphorus 0.8, and potassium 1.0. A coefficient fusion tool is used to multiply the leaching ratio by the corresponding nutrient leaching coefficient to obtain nutrient leaching sensitivity parameters. For example, the sensitivity parameter for urea (nitrogen) is 0.06 × 1.2 = 0.072, the sensitivity parameter for potassium chloride (potassium) is 0.06 × 1.0 = 0.06, and the sensitivity parameter for phosphate fertilizer (phosphorus) with a leaching ratio of 0.04 is 0.04 × 0.8 = 0.032. The parameter range is controlled between 0.02 and 0.1 to ensure that the sensitivity of different nutrients can be compared horizontally.

[0043] Furthermore, a leaching risk escalation coefficient is generated by multiplying soil moisture permeability parameters and nutrient leaching sensitivity parameters, and by dynamically adjusting the calculation by introducing future precipitation probabilities predicted based on meteorological impact datasets.

[0044] In this embodiment of the invention, a data multiplication tool is used to multiply the soil moisture permeability parameter and the nutrient leaching sensitivity parameter. Taking a soil moisture permeability parameter of 10.36 and a nitrogen sensitivity parameter of 0.072 as an example, the product result is 10.36 × 0.072 ≈ 0.746. Precipitation probability forecast data (in percentage) for the next 7 days is extracted from the meteorological impact dataset, and the average value is used as the adjustment basis. For example, if the precipitation probabilities for the next 7 days are 20%, 30%, 0%, 10%, 40%, 25%, and 15%, the average value is (20 + 30 + 0 + 10 + 40 + 25 + 15) ÷ 7 ≈ 20%. The precipitation probability adjustment formula is set as follows: adjustment coefficient = 1 + (precipitation probability ÷ 100) × 0.5, and the calculated adjustment coefficient is 1 + (20 ÷ 100) × 0.5 = 1.1. Multiply the product result with the adjustment coefficient to generate the leaching risk increase coefficient. For example, the nitrogen leaching risk increase coefficient = 0.746 × 1.1 ≈ 0.821. If the average probability of future precipitation is 50%, the adjustment coefficient = 1 + 0.5 × 0.5 = 1.25. The potassium product result is 0.621, and the leaching risk increase coefficient = 0.621 × 1.25 ≈ 0.776. The coefficient range is set to 0.5-1.5 to meet the calculation requirements of the subsequent nutrient leaching risk model.

[0045] Furthermore, step S3 includes the following steps: Step S31: Construct a crop growth-fertilizer response model based on the four-dimensional regulation matrix of agricultural crops, which includes a crop growth simulation module, a soil nutrient migration module, a meteorological impact simulation module, and a fertilizer effect calculation module. Among them, the crop growth simulation module is used to simulate the changes in crop physiological indicators under different nutrient supply, the soil nutrient migration module is used to simulate the transformation and loss of soil nutrients after fertilizer application, the meteorological impact simulation module is used to simulate the effects of light and precipitation on crop nutrient absorption and soil nutrient availability, and the fertilizer effect calculation module is used to quantify the nutrient supply efficiency of different fertilizer schemes. In this embodiment of the invention, a crop growth-fertilizer response model is built based on a four-dimensional crop regulation matrix (5 growth stages × 3 soil fertility levels × 3 meteorological conditions × 12 fertilizer schemes) using a model decomposition and construction tool. The model contains four core modules. The crop growth simulation module takes into account the nutrient supply parameters (e.g., nitrogen 80-180 mg / kg) from the four-dimensional matrix and simulates the daily changes in plant height, stem diameter, and chlorophyll SPAD value using the physiological index change formula (plant height = initial plant height + growth rate × number of growing days, where growth rate is linearly positively correlated with nutrient supply), outputting time-series data of crop physiological indicators. Soil nutrient migration module: Input fertilizer application parameters (e.g., urea 20 kg / mu) and soil parameters (e.g., permeability coefficient 2.2 mm / h). Using nutrient conversion formulas (organic nitrogen mineralization rate = 0.02 × temperature - 0.1, temperature taken from meteorological parameters) and leaching formulas (leaching amount = permeability coefficient × soil moisture content × nutrient concentration), the module simulates the daily changes in nitrogen, phosphorus, and potassium content in the 0-60 cm soil layer after fertilizer application and outputs soil nutrient migration curves. Meteorological impact simulation module: Input light duration (3-10 h) and precipitation (10-50 mm / week). Using light impact formulas (chlorophyll synthesis rate = 0.03 × light duration) and precipitation impact formulas (soil nutrient availability = 0.8 × moisture content + 0.2, moisture content calculated from precipitation and evaporation), the module quantifies the impact of meteorological factors on crop absorption and soil nutrient uptake and outputs time-series data of the impact coefficients. The fertilizer effect calculation module inputs the fertilizer type (urea, superphosphate, etc.) and application dosage. Using the supply efficiency formula (efficiency = total crop absorption / total fertilizer nutrients × 100%), it calculates the nutrient supply efficiency for different scenarios and outputs efficiency parameters. The four modules share data, forming a complete crop growth-fertilizer response model.

[0046] Step S32: Generate multiple fertilizer application scheme combinations through orthogonal experiments, including combinations of different fertilizer types and ratios, different application doses, and different application intervals. Combine the soil and meteorological parameters in the four-dimensional regulation matrix of agricultural crops, input each scheme into the crop growth-fertilizer response model for simulation calculation, and obtain the crop growth curve, soil nutrient change curve and crop nutrient absorption curve under different schemes. In this embodiment of the invention, an orthogonal experimental design tool was used to design 3×3×2=18 orthogonal experimental schemes based on fertilizer type ratio (urea: superphosphate: potassium chloride, with 3 levels: 1:1:0, 1:0:1, 2:1:1), application dosage (3 levels: 15 kg / mu, 20 kg / mu, 25 kg / mu), and application interval (2 levels: 15 days, 20 days). Each scheme had specific parameters, such as Scheme 1: ratio 1:1:0, dosage 15 kg / mu, interval 15 days; Scheme 8: ratio 2:1:1, dosage 20 kg / mu, interval 20 days. Typical soil parameters (loam, nitrogen 130 mg / kg, phosphorus 40 mg / kg, potassium 150 mg / kg) and meteorological parameters (average daily sunshine 7h, diurnal temperature range 10℃, weekly precipitation 30 mm) were extracted from the four-dimensional regulation matrix of agricultural crops as fixed input conditions. Using the model input tool, 18 schemes were input into the crop growth-fertilizer response model. The simulation period was set to 120 days (the complete growth cycle). Data was output once a day to obtain the crop growth curve (plant height and stem diameter changes with the number of days), soil nutrient change curve (nitrogen, phosphorus and potassium content changes with the number of days), and crop nutrient absorption curve (daily changes in nitrogen, phosphorus and potassium absorption) for each scheme. For example, the simulation of scheme 8 showed that the plant height increased by 0.8 cm per day during the jointing stage, the soil nitrogen content decreased from 130 mg / kg to 110 mg / kg, and the average daily nitrogen absorption was 0.5 kg / mu.

[0047] Step S33: Extract features from the simulated curves to calculate the peak crop growth rate, soil nutrient maintenance cycle, and crop nutrient absorption efficiency, and generate a fertilizer regulation simulation dataset. In this embodiment of the invention, the simulation curves of 18 schemes were analyzed using a curve feature extraction tool. For the crop growth curve (taking plant height as an example), the daily growth rate was calculated using the formula (growth rate = (current day's plant height - previous day's plant height) / 1 day). The maximum rate was selected as the peak growth rate. For example, in Scheme 8, the plant height curve showed a growth rate of 1.2 cm / day on day 45, which was the peak. For the soil nutrient change curve (taking nitrogen content as an example), the number of days the soil nitrogen content was maintained at 100-150 mg / kg (suitable range) was defined as the soil nutrient maintenance period. For example, in Scheme 8, the nitrogen content decreased from 130 mg / kg on day 10 to 98 mg / kg on day 50, with a maintenance period of 40 days. For the crop nutrient absorption curve (taking the nitrogen absorption curve as an example), the crop nutrient absorption efficiency is calculated using the formula (absorption efficiency = total absorption / (initial soil nitrogen content + fertilizer nitrogen content) × 100%). For example, in Scheme 8, the total nitrogen absorption is 12 kg / mu, the initial soil nitrogen is 130 mg / kg (approximately 26 kg / mu), and the fertilizer nitrogen is 20 kg / mu × 46% = 9.2 kg / mu. The absorption efficiency is 12 / (26 + 9.2) × 100% ≈ 34.1%. The peak growth rate (plant height, stem diameter), soil nutrient maintenance period (nitrogen, phosphorus, potassium), and crop nutrient absorption efficiency (nitrogen, phosphorus, potassium) of the 18 schemes are summarized to generate a fertilizer regulation simulation dataset containing 18 records and 8 indicators.

[0048] Step S34: Perform regulation coupling analysis based on fertilizer regulation simulation dataset and crop growth characteristic dataset to determine the initial fertilizer application scheme set corresponding to each growth stage to meet crop nutrient requirements and avoid soil nutrient imbalance.

[0049] In this embodiment of the invention, a regulation coupling analysis tool is used to perform correlation analysis between the fertilizer regulation simulation dataset and the crop growth characteristic dataset. Crop nutrient requirement thresholds (e.g., 5-7 kg / mu nitrogen requirement at the seedling stage, 8-10 kg / mu at the jointing stage) and soil nutrient balance thresholds (80-120 mg / kg nitrogen at the seedling stage, 90-150 mg / kg nitrogen at the jointing stage) for each growth stage are extracted from the crop growth characteristic dataset and used as the criteria for coupling analysis. For the 18 schemes in the fertilizer regulation simulation dataset, the data is split according to growth stage (seedling stage 1-30 days, jointing stage 31-60 days, fruiting stage 61-120 days), and each scheme is verified to ensure that the crop nutrient absorption reaches the requirement threshold range and the soil nutrient content remains within the balance threshold range at each stage. For example, Scheme 8 requires 6.5 kg / mu of nitrogen uptake during the seedling stage (meeting the requirement of 5-7 kg / mu) and a soil nitrogen maintenance period of 30 days (covering the seedling stage); 8.8 kg / mu of nitrogen uptake during the jointing stage (meeting the requirement of 8-10 kg / mu) and a soil nitrogen maintenance period of 40 days (covering the jointing stage); and 7.2 kg / mu of nitrogen uptake during the fruiting stage (assuming a requirement threshold of 6-8 kg / mu, which is met) and a soil nitrogen maintenance period of 35 days (covering the fruiting stage). Scheme 8 is therefore deemed to meet the requirements. All schemes are then screened, and those that meet the requirements for each stage, such as Schemes 1, 3, 8, 12, and 15, are retained to form an initial fertilizer application scheme set. Each scheme includes fertilizer type and ratio, application dosage, application interval, and the suitability assessment results for the corresponding stage.

[0050] Furthermore, step S34 includes the following steps: Step S341: Extract the total crop nutrient absorption, soil nutrient residue and fertilizer utilization efficiency parameters corresponding to each fertilizer scheme from the fertilizer regulation simulation dataset, and extract the crop nutrient demand threshold and soil nutrient balance threshold for each growth stage from the crop growth characteristic dataset to construct a simulation-actual parameter mapping table. In this embodiment of the invention, parameters corresponding to six fertilizer schemes (Scheme 1: urea 15 kg / mu + superphosphate 10 kg / mu; Scheme 2: urea 20 kg / mu + potassium chloride 12 kg / mu; Scheme 3: compound fertilizer 25 kg / mu; Scheme 4: urea 18 kg / mu + superphosphate 12 kg / mu + potassium chloride 8 kg / mu; Scheme 5: superphosphate 15 kg / mu + potassium chloride 10 kg / mu; Scheme 6: urea 22 kg / mu + compound fertilizer 15 kg / mu) were extracted from the fertilizer regulation simulation dataset using a simulation data extraction tool. These parameters included the total amount of nitrogen, phosphorus, and potassium absorbed by the crop (unit kg / mu), the residual amount of nitrogen, phosphorus, and potassium in the soil (unit mg / kg), and the fertilizer utilization efficiency (unit %). For example, in Scheme 2, the total nitrogen absorption at the jointing stage was 8.5 kg / mu, phosphorus was 0.6 kg / mu, and potassium was 5.2 kg / mu, with residual nitrogen in the soil of 140 mg / kg, phosphorus of 35 mg / kg, and potassium of 130 mg / kg, and a utilization efficiency of 68%. Extract crop nutrient requirement thresholds (e.g., nitrogen requirement threshold of 8-10 kg / mu, phosphorus 0.5-0.8 kg / mu, and potassium 5-7 kg / mu) and soil nutrient balance thresholds (same as the threshold for the jointing stage in the previous steps) from the crop growth characteristic dataset for the seedling stage, jointing stage, and fruiting stage. Use a mapping table construction tool to fill in the corresponding simulation parameters and actual requirement thresholds, with fertilizer schemes (6 types) as rows and growth stages (3 types) as columns, to form a 6×3 simulation-actual parameter mapping table.

[0051] Step S342: Based on the simulation-actual parameter mapping table, the ratio between the total amount of crop nutrient absorption and the corresponding crop nutrient demand threshold at each growth stage is used as the basis. Combined with the fertilizer utilization efficiency parameter, the corresponding nutrient absorption ratio in the crop growth characteristic dataset is also introduced to perform supply and demand matching calculation to obtain the crop nutrient supply and demand matching degree parameter. In this embodiment of the invention, crop nutrient supply and demand matching parameters are calculated using a supply and demand matching accounting tool based on a simulation-actual parameter mapping table. Taking the jointing stage of Scheme 2 as an example, the ratio of total crop nutrient absorption to the demand threshold is first calculated: nitrogen ratio = 8.5 kg / mu ÷ 9 kg / mu (median demand threshold) ≈ 0.94, phosphorus ratio = 0.6 kg / mu ÷ 0.65 kg / mu ≈ 0.92, potassium ratio = 5.2 kg / mu ÷ 6 kg / mu ≈ 0.87. The nutrient absorption ratios at the jointing stage (nitrogen 55%, phosphorus 15%, potassium 30%) are extracted from the crop growth characteristic dataset, and the basic matching degree is obtained by weighting each nutrient ratio according to the absorption ratio = 0.94 × 55% + 0.92 × 15% + 0.87 × 30% ≈ 0.92. Based on Scheme 2, where the fertilizer utilization efficiency at the jointing stage is 68%, the supply-demand matching degree parameter is calculated using the formula: Basic Matching Degree × (Utilization Efficiency ÷ 100). The result is 0.92 × 0.68 ≈ 0.625 (parameter range 0-1, the larger the value, the higher the matching degree). If Scheme 4 has a nitrogen ratio of 0.98, phosphorus 0.95, and potassium 0.93 at the jointing stage, with the same absorption ratio and a utilization efficiency of 75%, then the basic matching degree = 0.98 × 55% + 0.95 × 15% + 0.93 × 30% ≈ 0.96, and the supply-demand matching degree parameter = 0.96 × 0.75 = 0.72.

[0052] Step S343: Based on the soil nutrient residue and soil nutrient balance threshold corresponding to the simulation-actual parameter mapping table, perform reverse mapping of soil nutrient imbalance to generate soil nutrient balance parameters. In this embodiment of the invention, soil nutrient balance parameters are generated using an imbalance inverse mapping tool based on soil nutrient residues and soil nutrient balance thresholds within a simulation-actual parameter mapping table. Taking the jointing stage of Scheme 2 as an example, the calculation is performed according to the S3431-S3435 process: First, a three-dimensional correlation table of nutrient type-residue-threshold is constructed (nitrogen residue 140 mg / kg, threshold 90-150 mg / kg; phosphorus 35 mg / kg, threshold 25-40 mg / kg; potassium 130 mg / kg, threshold 110-140 mg / kg). The individual nutrient imbalance difference values ​​are calculated (nitrogen 0, phosphorus 0, potassium 0). Combined with soil buffering capacity parameters (nitrogen 1.2, phosphorus 1.5, potassium 1.1), the corrected risk values ​​(all 0) are obtained. The overall imbalance risk coefficient is 0. Substituting this into the inverse mapping model (balance degree = 100 - risk coefficient × 5), the initial balance degree is 100. The nutrient uptake rates of crops at the jointing stage were extracted (nitrogen 2.5 mg / kg·day, phosphorus 1.2 mg / kg·day, potassium 1.8 mg / kg·day), and the uptake-residue matching coefficient was calculated as (2.5+1.2+1.8) / (140+35+130)×10≈0.18. The final soil nutrient balance parameter was 100×0.18=18 (parameter range 0-100, the larger the value, the higher the balance).

[0053] Step S344: Weight and integrate the crop nutrient supply and demand matching parameters with the soil nutrient balance parameters to generate a comprehensive evaluation score for each fertilizer program; In this embodiment of the invention, a weighted fusion tool is used to integrate the crop nutrient supply and demand matching parameter with the soil nutrient balance parameter, setting the weight of supply and demand matching at 0.6 and the weight of soil balance at 0.4 (because supply and demand matching directly affects crop growth, the weight is higher). Taking the jointing stage of Scheme 2 as an example, the supply and demand matching parameter is 0.625 (converted to 62.5 points) and the soil balance parameter is 18. According to the formula, the comprehensive evaluation score is (62.5 × 0.6) + (18 × 0.4) = 37.5 + 7.2 = 44.7 points (the score range is 0-100, and the higher the value, the better the scheme). Scheme 4: Supply and demand matching parameter at the jointing stage is 0.72 (72 points), soil balance parameter is 22, and the comprehensive evaluation score is 72×0.6+22×0.4=43.2+8.8=52 points; Scheme 6: Supply and demand matching parameter at the jointing stage is 0.58 (58 points), soil balance parameter is 15, and the comprehensive evaluation score is 58×0.6+15×0.4=34.8+6=40.8 points.

[0054] Step S345: Based on the comprehensive evaluation score of each fertilizer scheme, the initial fertilizer application scheme set is screened. By sorting the comprehensive evaluation scores from high to low, fertilizer schemes with scores higher than a preset threshold are selected as candidate schemes. Redundancy analysis is performed on the candidate schemes to calculate the difference in fertilizer parameters between different schemes. Based on the difference in fertilizer parameters between different schemes, the initial fertilizer application scheme set that meets the crop nutrient requirements at each growth stage and avoids soil nutrient imbalance is determined. Each scheme includes fertilizer type ratio, application dosage and application cycle.

[0055] In this embodiment of the invention, a preset threshold of 45 points for the comprehensive evaluation score is set. A scheme ranking tool is used to sort the scores of six schemes at the jointing stage from highest to lowest: Scheme 4 (52 points), Scheme 2 (44.7 points, below the threshold), Scheme 6 (40.8 points), etc., and Scheme 4 is selected as a candidate scheme. Redundancy analysis is performed on all candidate schemes at all growth stages. A parameter difference calculation tool is used to calculate the difference in fertilizer parameters between different schemes using the formula: Difference = 1 - (|Scheme A fertilizer dosage - Scheme B fertilizer dosage| / max(Scheme A dosage, Scheme B dosage)). For example, the difference between Scheme 4 (urea 18 kg / mu) and Scheme 7 (urea 17 kg / mu) is 1 - (1 / 18) ≈ 0.94. A difference below 0.1 is considered redundant. Based on the difference screening, non-redundant schemes were selected, and the initial fertilizer application scheme set for each growth stage was determined: seedling stage scheme 3 (compound fertilizer 25kg / mu, score 48 points), jointing stage scheme 4 (urea 18kg / mu + superphosphate 12kg / mu + potassium chloride 8kg / mu, score 52 points), and fruiting stage scheme 1 (urea 15kg / mu + superphosphate 10kg / mu, score 46 points). Each scheme specifies the fertilizer type ratio, application dosage, and application cycle for the corresponding growth stage (11 days of seedling stage, 31 days of jointing stage, and 91 days of fruiting stage).

[0056] Furthermore, step S343 includes the following steps: Soil nutrient residues corresponding to each fertilizer scheme are extracted from the simulation-actual parameter mapping table. At the same time, soil nutrient balance thresholds corresponding to the growth stage and soil type are extracted to construct a three-dimensional correlation table of nutrient type-residue-threshold. In this embodiment of the invention, the soil nitrogen, phosphorus, and potassium residues (unit: mg / kg) for four fertilizer schemes (urea 15 kg / mu, urea 20 kg / mu, superphosphate 15 kg / mu, and potassium chloride 15 kg / mu) at the seedling stage, jointing stage, and fruiting stage are extracted from the simulation-actual parameter mapping table using a mapping table data extraction tool. For example, at the jointing stage, the nitrogen residue of urea 20 kg / mu is 140 mg / kg, the phosphorus residue is 35 mg / kg, and the potassium residue is 130 mg / kg. Simultaneously, soil nutrient balance thresholds corresponding to the growth stage and soil type (loam) were extracted from the soil environmental characteristic dataset. Seedling stage nitrogen balance thresholds were set with an upper limit of 120 mg / kg and a lower limit of 80 mg / kg, phosphorus upper limit of 30 mg / kg and a lower limit of 20 mg / kg, and potassium upper limit of 120 mg / kg and a lower limit of 100 mg / kg. Jointing stage nitrogen upper limit was set at 150 mg / kg and a lower limit at 90 mg / kg, phosphorus upper limit at 40 mg / kg and a lower limit of 25 mg / kg, and potassium upper limit at 140 mg / kg and a lower limit of 110 mg / kg. A three-dimensional correlation table was constructed using nutrient type (nitrogen, phosphorus, potassium) as rows, fertilizer schemes (4 types) as columns, and growth stages (3 stages) as depth. Corresponding residue levels and balance thresholds were entered to form a 3×4×3 nutrient type-residue-threshold three-dimensional correlation table.

[0057] Furthermore, based on the three-dimensional correlation table of nutrient type-residue-threshold, the residue-threshold difference parameter for each nutrient type is calculated. For each nutrient, the positive difference between residue and upper threshold is calculated separately. If the residue exceeds the upper threshold, it is a positive value; otherwise, it is 0. The negative difference between lower threshold and residue is calculated separately. If the residue is below the lower threshold, it is a positive value; otherwise, it is 0. The positive and negative differences are weighted and summed to generate a single nutrient imbalance difference value. In this embodiment of the invention, the residue-threshold difference parameters for each nutrient type are calculated using a difference parameter calculation tool based on a three-dimensional correlation table of nutrient type-residue-threshold. Taking a urea application of 20 kg / mu at the jointing stage as an example, the nitrogen residue is 140 mg / kg, with an upper limit of 150 mg / kg, resulting in a positive difference of 140 - 150 = -10 mg / kg (set to 0, as it does not exceed the upper limit); the nitrogen lower limit is 90 mg / kg, resulting in a negative difference of 90 - 140 = -50 mg / kg (set to 0, as it is not below the lower limit). The phosphorus residue is 35 mg / kg, with an upper limit of 40 mg / kg, resulting in a positive difference of 35 - 40 = -5 mg / kg (set to 0); the phosphorus lower limit is 25 mg / kg, resulting in a negative difference of 25 - 35 = -10 mg / kg (set to 0). Potassium residue is 130 mg / kg, upper limit is 140 mg / kg, positive difference = 130 - 140 = -10 mg / kg (taken as 0); potassium lower limit is 110 mg / kg, negative difference = 110 - 130 = -20 mg / kg (taken as 0). Setting a weight of 0.6 for positive difference and 0.4 for negative difference, the weighted sum is used to generate the single nutrient imbalance difference value. For example, under this scheme, nitrogen imbalance difference value = 0 × 0.6 + 0 × 0.4 = 0, phosphorus = 0, potassium = 0. If the nitrogen residue in a certain scheme is 160 mg / kg (exceeding the upper limit of 150 mg / kg), the positive difference is 10 mg / kg, the negative difference is 0, and the imbalance difference is 10 × 0.6 + 0 × 0.4 = 6; if the nitrogen residue is 70 mg / kg (below the lower limit of 80 mg / kg), the negative difference is 10 mg / kg, the positive difference is 0, and the imbalance difference is 0 × 0.6 + 10 × 0.4 = 4.

[0058] Furthermore, the difference value of single nutrient imbalance is corrected by introducing the soil nutrient buffering capacity parameter from the soil environmental characteristic dataset. The difference value of single nutrient imbalance is multiplied by the reciprocal of the soil nutrient buffering capacity parameter to obtain the corrected risk value of single nutrient imbalance. The corrected risk values ​​of single nutrient imbalance for all nutrient types are summed to generate the comprehensive soil nutrient imbalance risk coefficient. In this embodiment of the invention, soil nutrient buffering capacity parameters for the corresponding soil type (loam) are extracted from the soil environmental characteristic dataset, and nitrogen buffering capacity is set to 1.2, phosphorus to 1.5, and potassium to 1.1 (parameter range 1.0-2.0, the larger the value, the stronger the buffering capacity). A risk value correction tool is used, and the risk value of a single nutrient imbalance is calculated according to the formula: Risk value of single nutrient imbalance = Difference value of single nutrient imbalance × (1 / Soil nutrient buffering capacity parameter). For example, in a certain scheme, with a nitrogen imbalance difference value of 6 and a buffering capacity of 1.2, the risk value = 6 × (1 / 1.2) = 5; with a phosphorus imbalance difference value of 4 and a buffering capacity of 1.5, the risk value = 4 × (1 / 1.5) ≈ 2.67; and with a potassium imbalance difference value of 3 and a buffering capacity of 1.1, the risk value = 3 × (1 / 1.1) ≈ 2.73. Using the risk coefficient summation tool, the corrected risk values ​​for all nutrient types are added together to generate the comprehensive soil nutrient imbalance risk coefficient = 5 + 2.67 + 2.73 = 10.4 (the coefficient ranges from 0 to 20, and the larger the value, the higher the risk of imbalance).

[0059] Furthermore, an imbalance risk-balance degree inverse mapping model is constructed to set the value range of the comprehensive soil nutrient imbalance risk coefficient, establish the inverse mapping relationship between this range and the soil nutrient balance degree parameter, and substitute the generated comprehensive soil nutrient imbalance risk coefficient into the model to obtain the initial soil nutrient balance degree parameter. In this embodiment of the invention, an imbalance risk-balance inverse mapping model is constructed using a model building tool. The comprehensive soil nutrient imbalance risk coefficient is set to a range of 0-20, corresponding to a soil nutrient balance parameter range of 100-0 (balance 100 represents complete balance, 0 represents severe imbalance). A linear inverse mapping relationship is established: soil nutrient balance parameter = 100 - (comprehensive risk coefficient × 5). The generated comprehensive soil nutrient imbalance risk coefficient is substituted into the model to calculate the initial balance parameter. For example, when the coefficient is 10.4, the initial balance = 100 - (10.4 × 5) = 48; when the coefficient is 5, the initial balance = 100 - 25 = 75; when the coefficient is 0, the initial balance = 100. If the calculation result exceeds the range of 0-100, a boundary value is taken. For example, when the coefficient is 25, the initial balance = 0 (taking the lower limit), ensuring that the balance parameter can quantitatively assess the soil nutrient status.

[0060] Furthermore, the initial soil nutrient balance parameter is dynamically optimized by combining the crop nutrient absorption rate in the crop growth characteristic dataset to calculate the ratio of crop nutrient absorption rate to soil nutrient residue, generate the absorption-residue matching coefficient, and multiply the initial soil nutrient balance parameter with the absorption-residue matching coefficient to generate the soil nutrient balance parameter.

[0061] In this embodiment of the invention, the crop nutrient absorption rate (unit: mg / kg·day) corresponding to the growth stage is extracted from the crop growth characteristic dataset. For example, the nitrogen absorption rate at the jointing stage is 2.5 mg / kg·day, phosphorus is 1.2 mg / kg·day, and potassium is 1.8 mg / kg·day. A matching coefficient calculation tool is used, and the absorption-residue matching coefficient is calculated using the formula: absorption-residue matching coefficient = (total crop nutrient absorption rate) / (total soil nutrient residue) × 10. For example, in a certain scheme, the nitrogen residue is 140 mg / kg, phosphorus is 35 mg / kg, and potassium is 130 mg / kg. The total absorption rate is 2.5 + 1.2 + 1.8 = 5.5 mg / kg·day, and the total residue is 140 + 35 + 130 = 305 mg / kg. The matching coefficient is (5.5 / 305) × 10 ≈ 0.18. Using a balance optimization tool, the initial soil nutrient balance parameter is multiplied by the absorption-residue matching coefficient to generate the final soil nutrient balance parameter. For example, if the initial balance is 48 and the matching coefficient is 0.18, the final balance is 48 × 0.18 ≈ 8.6; if the matching coefficient is 0.3 and the initial balance is 75, the final balance is 75 × 0.3 = 22.5. This achieves dynamic optimization of the balance assessment results based on crop absorption.

[0062] Furthermore, step S4 includes the following steps: Step S41: Design a dynamic adjustment mechanism based on the initial fertilizer application scheme set, including the classification of control levels, deviation triggering conditions, and model recalibration rules. In this embodiment of the invention, a dynamic adjustment mechanism is designed based on an initial fertilizer application scheme set (schemes 1, 3, 8, 12, and 15) using a mechanism design tool. The control levels are divided into three levels according to the magnitude of the fertilizer control deviation parameters: Level 1 deviation (0-5), Level 2 deviation (6-10), and Level 3 deviation (>10), corresponding to adjustment ranges of ±5%, ±10%, and ±15% of the current dosage, respectively. Deviation triggering conditions are set for key indicators (plant height, chlorophyll SPAD value, and soil nitrogen content) at each growth stage. For seedlings, the deviation thresholds are ±1cm for plant height, ±3 for chlorophyll, and ±10mg / kg for soil nitrogen. For jointing stage, the thresholds are ±2cm for plant height, ±4 for chlorophyll, and ±15mg / kg for soil nitrogen. Adjustment is triggered when any indicator deviation exceeds the threshold. Model recalibration rules: Every 30 days, crop growth-fertilizer response model parameters are calibrated based on real-time data, such as the leaching formula coefficients of the soil nutrient migration module. When the actual leaching amount deviates from the simulated value by more than 10%, the coefficients are recalculated (new coefficient = original coefficient × actual leaching amount / simulated leaching amount) to ensure the model's prediction accuracy. Control levels, trigger conditions, and calibration rules are integrated into an executable dynamic adjustment mechanism document.

[0063] Step S42: Obtain real-time growth status data of agricultural crops through real-time monitoring equipment, compare the real-time growth status data with the simulation data of the corresponding growth stage in the fertilizer regulation simulation dataset, calculate the deviation value of each growth indicator, and calculate the fertilizer regulation deviation parameter by combining the regulation weight of each indicator in the four-dimensional regulation matrix of agricultural crops. In this embodiment of the invention, real-time growth status data on the 40th day of the jointing stage is obtained through real-time monitoring equipment (growth monitor, soil analyzer): plant height 65cm, chlorophyll SPAD value 45, soil nitrogen content 125mg / kg. Simulation data on the 40th day of the jointing stage of scheme 8 is extracted from the fertilizer regulation simulation dataset: plant height 67cm, chlorophyll 48, soil nitrogen 135mg / kg. Using a deviation calculation tool, the deviation values ​​of each indicator are calculated according to the formula (real-time value - simulation value): plant height -2cm, chlorophyll -3, soil nitrogen -10mg / kg. The regulation weights of each indicator are extracted from the four-dimensional regulation matrix of agricultural crops: plant height 0.3, chlorophyll 0.4, soil nitrogen 0.3. The fertilizer regulation deviation parameter is calculated according to the formula (deviation parameter = Σ|indicator deviation value| × weight) = (2×0.3) + (3×0.4) + (10×0.3) = 0.6 + 1.2 + 3 = 4.8.

[0064] Step S43: Determine whether to trigger the dynamic adjustment mechanism based on the fertilizer regulation deviation parameter. If the deviation parameter is within the preset acceptable range, maintain the current fertilizer application plan. If the deviation parameter exceeds the acceptable range, match the corresponding regulation level according to the deviation size and start the fertilizer plan re-optimization process corresponding to the crop growth-fertilizer response model. In this embodiment of the invention, by employing a deviation determination tool, the calculated fertilizer regulation deviation parameter 4.8 is compared with the preset acceptable range (0-5). If it is determined to be within the level 1 deviation range, no adjustment is required, and the current scheme 8 (urea 18kg / mu + superphosphate 12kg / mu + potassium chloride 8kg / mu) is maintained. If the real-time data is plant height 63cm, chlorophyll 42, and soil nitrogen 115mg / kg, with deviation values ​​of -4cm, -6, and -20mg / kg respectively, the deviation parameter = (4×0.3) + (6×0.4) + (20×0.3) = 1.2 + 2.4 + 6 = 9.6, which exceeds the acceptable range (0-5), matching a level 2 deviation, and initiating the fertilizer scheme re-optimization process of the crop growth-fertilizer response model.

[0065] Step S44: In the fertilizer scheme re-optimization process, with the goal of reducing fertilizer regulation deviation parameters, the fertilizer effect parameters and soil nutrient migration parameters in the crop growth-fertilizer response model are dynamically adjusted to regenerate the fertilizer regulation simulation dataset; based on the new fertilizer regulation simulation dataset and combined with real-time soil environmental data and meteorological forecast data, the optimal fertilizer regulation scheme for each growth stage is output. In this embodiment of the invention, by employing parameter adjustment tools in the scheme re-optimization process, the parameters of the crop growth-fertilizer response model are adjusted with the goal of reducing the deviation parameter (targeting a value below 5): the urea supply efficiency parameter of the fertilizer effect calculation module is increased from 68% to 72% (because the actual absorption efficiency is lower than the simulation), and the nitrogen leaching coefficient of the soil nutrient migration module is decreased from 0.05 to 0.04 (to reduce leaching). The model is then rerun to generate a new fertilizer regulation simulation dataset. The new dataset shows the simulation data on day 40 of the jointing stage after the adjustment of Scheme 8 (urea 19.8 kg / mu, superphosphate 13.2 kg / mu, potassium chloride 8.8 kg / mu): plant height 64 cm, chlorophyll 44, soil nitrogen 120 mg / kg. Combining real-time soil environmental data (moisture content 22%, pH 6.5) and meteorological forecast data (20% probability of precipitation in the next 7 days, 8 hours of sunshine / day), a scheme optimization tool was used to screen the scheme with the smallest central deviation of the new data and output the optimal fertilizer regulation scheme for the jointing stage: urea 19.8 kg / mu, superphosphate 13.2 kg / mu, potassium chloride 8.8 kg / mu, with an application interval of 20 days.

[0066] Step S45: Transform the optimal fertilizer regulation plan into control instructions that the fertilizer application execution system can recognize and push them to the field fertilization equipment simultaneously; at the same time, display the crop growth prediction comparison, soil nutrient change prediction, and fertilizer utilization efficiency change before and after the plan adjustment through a visualization platform, and generate a dynamic report on fertilizer regulation.

[0067] In this embodiment of the invention, an instruction conversion tool is used to convert the optimal fertilizer regulation scheme into control instructions that the fertilization execution system can recognize. These instructions include fertilizer type (urea, superphosphate, potassium chloride), dosage (19.8 kg / mu, 13.2 kg / mu, 8.8 kg / mu), application time (41 days after the jointing stage), and operating parameters of the fertilization equipment (spreading speed 2 m / s, row spacing 5 m), which are simultaneously pushed to the field fertilization equipment (intelligent fertilizer spreader, drip irrigation system). A visualization platform is used to generate charts to display the comparison before and after the adjustment: crop growth prediction comparison (plant height 67 cm before adjustment → 64 cm after adjustment, closer to the real-time 63 cm), soil nutrient change prediction (135 mg / kg before adjustment → 120 mg / kg after adjustment, close to the real-time 115 mg / kg), and fertilizer utilization efficiency change (68% before adjustment → 72% after adjustment). By integrating the comparative data, optimal scheme parameters, and execution instructions, a dynamic fertilizer regulation report is generated. The report includes the reasons for the adjustment, scheme details, and expected effects for subsequent regulation reference.

[0068] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0069] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for intelligent regulation of fertilizers during the growth stages of agricultural crops based on multiple factors, characterized in that, Includes the following steps: Step S1: Obtain the growth status data, soil environment data, meteorological data and fertilizer application history data of agricultural crops, and perform multi-source heterogeneous data fusion and standardization processing to generate standardized crop growth characteristic datasets, soil environment characteristic datasets, meteorological impact datasets and fertilizer application characteristic datasets. Step S2: Extract the physiological index change patterns of crops at different growth stages based on crop growth characteristic datasets, and explore the correlation between soil fertility and crop nutrient absorption efficiency by combining soil environmental characteristic datasets. At the same time, introduce meteorological impact datasets and fertilizer application characteristic datasets to construct a four-dimensional regulation matrix for agricultural crops. Step S3: Construct a crop growth-fertilizer response model based on the four-dimensional regulation matrix of agricultural crops, and simulate the crop growth process under different fertilizer application schemes based on the crop growth-fertilizer response model to generate a fertilizer regulation simulation dataset; perform regulation coupling analysis based on the fertilizer regulation simulation dataset and the crop growth characteristic dataset to determine the initial fertilizer application scheme set corresponding to each growth stage that meets the crop nutrient requirements and avoids soil nutrient imbalance. Step S4: Design a dynamic adjustment mechanism based on the initial fertilizer application scheme set to compare and analyze the real-time crop growth status data with the fertilizer regulation simulation dataset to generate fertilizer regulation deviation parameters; drive the crop growth-fertilizer response model based on the fertilizer regulation deviation parameters to re-optimize the fertilizer scheme, so as to output the optimal fertilizer regulation scheme corresponding to each growth stage and synchronize it to the fertilization execution system, thereby realizing the dynamic visualization and execution of agricultural crop fertilizer regulation.

2. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Collect growth status data of agricultural crops, including plant height, stem diameter, number of leaves, leaf area, and chlorophyll SPAD value, through crop growth monitoring equipment. Step S12: Collect soil environmental data including soil nitrogen, phosphorus and potassium content, soil pH value, soil moisture content and soil bulk density using soil testing equipment; Step S13: Collect meteorological data corresponding to the average daily sunshine duration, cumulative precipitation, diurnal temperature range, and frequency of extreme weather events in agricultural crop planting areas by combining meteorological stations and satellite remote sensing. Step S14: Extract fertilizer application records from the agricultural production management system for the historical planting cycle of agricultural crops, including fertilizer type, dosage for each application, application time, and application method corresponding to historical fertilizer application data. Step S15: Filter outliers and complete the time series data for growth status data, perform spatial interpolation and standardization on soil environmental data, align the time series data and quantify the degree of influence on meteorological data, and encode the features of historical fertilizer application data to generate standardized crop growth characteristic datasets, soil environmental characteristic datasets, meteorological influence datasets, and fertilizer application characteristic datasets.

3. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Extract time-series data of physiological indicators corresponding to each growth stage of the crop, including sowing period, seedling stage, jointing stage, flowering stage and fruiting stage, from the crop growth characteristic dataset, and calculate the growth rate and coefficient of variation of the physiological indicators of each stage to generate a crop growth stage feature vector. Step S22: Based on the soil environmental feature dataset, analyze the correlation between soil nitrogen, phosphorus and potassium content and crop chlorophyll content and leaf nitrogen content, and generate soil fertility-nutrient absorption correlation parameters by calculating the contribution coefficient of soil nutrients to crop nutrient absorption; at the same time, calculate the mapping relationship between soil pH value and soil nutrient availability and combine it with the soil fertility-nutrient absorption correlation parameters to generate soil fertility-nutrient feature vector. Step S23: Calculate the correlation between sunshine duration, temperature, precipitation and crop growth rate in the meteorological impact dataset to obtain the influence weight coefficient of each meteorological factor on crop growth; Based on the influence weight coefficient of each meteorological factor on crop growth and combined with the water and light requirements of crops at different growth stages, generate meteorological factor-growth response correlation parameters. Step S24: Extract crop yield and quality indicators corresponding to different fertilizer types and application doses from the fertilizer application feature dataset, calculate fertilizer utilization efficiency, and generate fertilizer effect parameters. Step S25: Perform dimensional matching on the crop growth stage feature vector, soil fertility-nutrient feature vector, meteorological factor-growth response correlation parameters, and fertilizer effect parameters to construct a four-dimensional regulation matrix for agricultural crops. Each element in this matrix represents the contribution of the combination of soil, meteorological conditions, and fertilizer schemes to the regulation of crop growth at a certain growth stage.

4. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 3, characterized in that, Step S22 includes the following steps: Step S221: Extract time-series data of soil nitrogen, phosphorus and potassium content for multiple consecutive growth cycles from the soil environmental characteristics dataset, and extract nutrient absorption-related indicators for the corresponding cycles, including crop chlorophyll SPAD value, leaf nitrogen content and crop biomass, from the crop growth characteristics dataset. Step S222: Using time-series data of soil nitrogen, phosphorus and potassium content as independent variables and nutrient absorption-related indicators as dependent variables, construct a multiple linear regression model and calculate the model regression coefficients. The model regression coefficients are defined as the contribution coefficients of soil fertility to nutrient absorption. Step S223: Extract soil moisture content from the soil environmental characteristic dataset and analyze the impact of soil moisture content on soil nutrient dissolution and crop root absorption. When the soil moisture content is too low, calculate the decrease in nutrient dissolution rate; when the soil moisture content is too high, calculate the nutrient leaching risk coefficient. At the same time, use the decrease in nutrient dissolution rate and the nutrient leaching risk coefficient as correction terms to correct the basic contribution coefficient of soil fertility-nutrient absorption, and obtain the soil fertility-nutrient absorption correlation parameters. Step S224: Extract soil pH values ​​from the soil environmental characteristic dataset, and establish a mapping relationship table between soil pH values ​​and the availability of different nutrients by combining crop nutrient absorption characteristics; Step S225: Calculate the nutrient availability coefficient under different pH ranges based on the mapping relationship table, and multiply the nutrient availability coefficient by the soil nutrient content to obtain the soil available nutrient content; generate soil acid-base adjustment parameters based on the deviation between the soil available nutrient content and the actual absorption by crops; and generate soil fertility-nutrient feature vector by combining the soil fertility-nutrient absorption correlation parameters.

5. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 4, characterized in that, Step S223 includes the following steps: Real-time monitoring data of soil moisture content is extracted from the soil environmental characteristics dataset, and combined with the suitable moisture content range of the crop growth stage, the deviation value between the actual moisture content and the suitable moisture content is calculated to obtain the moisture content deviation parameter. A nutrient solubility model is constructed based on the moisture content deviation parameter. When the real-time soil moisture content is lower than or equal to the suitable moisture content range, the nutrient solubility decreases linearly with decreasing moisture content, and the decrease in nutrient solubility is calculated as (suitable moisture content - actual moisture content) / suitable moisture content × solubility decrease coefficient. When the real-time soil moisture content is higher than the suitable moisture content range, the nutrient leaching risk increases exponentially with increasing moisture content, and the nutrient leaching risk coefficient is calculated as (actual moisture content - suitable moisture content) / suitable moisture content × leaching risk increase coefficient. The decrease in nutrient solubility rate was weighted and combined with the nutrient leaching risk coefficient to obtain the comprehensive coefficient of soil moisture impact. The soil fertility-nutrient absorption baseline contribution coefficient is multiplied by the soil moisture influence comprehensive coefficient to obtain the corrected soil fertility-nutrient absorption correlation parameters. By comparing the differences in the correlation parameters before and after the correction, the correction effect is verified. If the degree of fit between the corrected parameters and the actual nutrient absorption data of the crop exceeds the preset threshold, the corrected soil fertility-nutrient absorption correlation parameters are retained; otherwise, the solubility reduction coefficient and the leaching risk increase coefficient are readjusted until the degree of fit meets the requirements.

6. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 5, characterized in that, The calculation process for the solubility reduction coefficient and the leaching risk increase coefficient includes the following steps: By extracting soil texture type, soil organic matter content and soil porosity data from soil environmental feature datasets, and combining them with the nutrient requirements of crop growth stages, a soil-crop nutrient compatibility assessment matrix is ​​constructed. Based on the soil-crop nutrient compatibility assessment matrix, the adsorption capacity parameters of different soil textures for nutrients at each growth stage are calculated. The soil nutrient adsorption capacity parameters are obtained by estimating and standardizing the product of soil organic matter content and soil porosity, combined with the corresponding adsorption capacity parameters. Extract historical fertilizer types and corresponding nutrient dissolution rate data from fertilizer application feature datasets, calculate the dissolution baseline coefficients of different fertilizer types in the current soil texture, and generate fertilizer-soil dissolution fit coefficients by combining soil nutrient adsorption capacity parameters; perform weighted summation based on adsorption capacity parameters and fertilizer-soil dissolution fit coefficients, and generate dissolution rate reduction coefficients by inverse exponential correction by introducing the concentration of organic acids secreted by crop roots. Soil saturated moisture content, soil permeability coefficient and groundwater depth data were extracted from the soil environmental characteristics dataset. A soil moisture leakage risk assessment model was constructed by combining historical precipitation intensity and duration data from the meteorological impact dataset. Based on the soil moisture leakage risk assessment model, the soil moisture leakage capacity parameter is estimated by the ratio of soil saturated moisture content to soil permeability coefficient and the leakage buffer coefficient corresponding to the groundwater level depth. Historical fertilizer leaching records are extracted from the fertilizer application characteristic dataset. Based on the historical fertilizer leaching records, the nutrient leaching sensitivity parameter is calculated by the ratio of historical leaching amount to corresponding fertilizer application amount and the leaching coefficient of nutrient type. The leaching risk escalation coefficient is generated by multiplying soil moisture permeability parameters and nutrient leaching sensitivity parameters, and by dynamically adjusting the calculation by introducing future precipitation probabilities predicted based on meteorological impact datasets.

7. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Construct a crop growth-fertilizer response model based on the four-dimensional regulation matrix of agricultural crops, which includes a crop growth simulation module, a soil nutrient migration module, a meteorological impact simulation module, and a fertilizer effect calculation module. Among them, the crop growth simulation module is used to simulate the changes in crop physiological indicators under different nutrient supply, the soil nutrient migration module is used to simulate the transformation and loss of soil nutrients after fertilizer application, the meteorological impact simulation module is used to simulate the effects of light and precipitation on crop nutrient absorption and soil nutrient availability, and the fertilizer effect calculation module is used to quantify the nutrient supply efficiency of different fertilizer schemes. Step S32: Generate multiple fertilizer application scheme combinations through orthogonal experiments, including combinations of different fertilizer types and ratios, different application doses, and different application intervals. Combine the soil and meteorological parameters in the four-dimensional regulation matrix of agricultural crops, input each scheme into the crop growth-fertilizer response model for simulation calculation, and obtain the crop growth curve, soil nutrient change curve and crop nutrient absorption curve under different schemes. Step S33: Extract features from the simulated curves to calculate the peak crop growth rate, soil nutrient maintenance cycle, and crop nutrient absorption efficiency, and generate a fertilizer regulation simulation dataset. Step S34: Perform regulation coupling analysis based on fertilizer regulation simulation dataset and crop growth characteristic dataset to determine the initial fertilizer application scheme set corresponding to each growth stage to meet crop nutrient requirements and avoid soil nutrient imbalance.

8. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 7, characterized in that, Step S34 includes the following steps: Step S341: Extract the total crop nutrient absorption, soil nutrient residue and fertilizer utilization efficiency parameters corresponding to each fertilizer scheme from the fertilizer regulation simulation dataset, and extract the crop nutrient demand threshold and soil nutrient balance threshold for each growth stage from the crop growth characteristic dataset to construct a simulation-actual parameter mapping table. Step S342: Based on the simulation-actual parameter mapping table, the ratio between the total amount of crop nutrient absorption and the corresponding crop nutrient demand threshold at each growth stage is used as the basis. Combined with the fertilizer utilization efficiency parameter, the corresponding nutrient absorption ratio in the crop growth characteristic dataset is also introduced to perform supply and demand matching calculation to obtain the crop nutrient supply and demand matching degree parameter. Step S343: Based on the soil nutrient residue and soil nutrient balance threshold corresponding to the simulation-actual parameter mapping table, perform reverse mapping of soil nutrient imbalance to generate soil nutrient balance parameters. Step S344: Weight and integrate the crop nutrient supply and demand matching parameters with the soil nutrient balance parameters to generate a comprehensive evaluation score for each fertilizer program; Step S345: Based on the comprehensive evaluation score of each fertilizer scheme, the initial fertilizer application scheme set is screened. By sorting the comprehensive evaluation scores from high to low, fertilizer schemes with scores higher than a preset threshold are selected as candidate schemes. Redundancy analysis is performed on the candidate schemes to calculate the difference in fertilizer parameters between different schemes. Based on the difference in fertilizer parameters between different schemes, the initial fertilizer application scheme set that meets the crop nutrient requirements at each growth stage and avoids soil nutrient imbalance is determined. Each scheme includes fertilizer type ratio, application dosage and application cycle.

9. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 8, characterized in that, Step S343 includes the following steps: Soil nutrient residues corresponding to each fertilizer scheme are extracted from the simulation-actual parameter mapping table. At the same time, soil nutrient balance thresholds corresponding to the growth stage and soil type are extracted to construct a three-dimensional correlation table of nutrient type-residue-threshold. The residual-threshold difference parameter for each nutrient type is calculated based on the three-dimensional correlation table of nutrient type-residue-threshold. For each nutrient, the positive difference between residual amount and upper threshold is calculated separately. If the residual amount exceeds the upper threshold, it is a positive value; otherwise, it is 0. The negative difference between lower threshold and residual amount is calculated separately. If the residual amount is lower than the lower threshold, it is a positive value; otherwise, it is 0. The positive and negative differences are weighted and summed to generate the single nutrient imbalance difference value. The individual nutrient imbalance difference value is corrected by introducing the soil nutrient buffering capacity parameter from the soil environmental characteristic dataset. The individual nutrient imbalance difference value is multiplied by the reciprocal of the soil nutrient buffering capacity parameter to obtain the corrected individual nutrient imbalance risk value. The corrected individual nutrient imbalance risk values ​​for all nutrient types are summed to generate the comprehensive soil nutrient imbalance risk coefficient. An inverse mapping model of imbalance risk and balance is constructed to set the range of values ​​for the comprehensive soil nutrient imbalance risk coefficient, establish the inverse mapping relationship between this range and the soil nutrient balance parameter, and substitute the generated comprehensive soil nutrient imbalance risk coefficient into the model to obtain the initial soil nutrient balance parameter. The initial soil nutrient balance parameter is dynamically optimized by combining the crop nutrient absorption rate in the crop growth characteristic dataset to calculate the ratio of crop nutrient absorption rate to soil nutrient residue, generate the absorption-residue matching coefficient, and multiply the initial soil nutrient balance parameter with the absorption-residue matching coefficient to generate the soil nutrient balance parameter.

10. The intelligent fertilizer regulation method for agricultural crop growth stages based on multiple factors according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: Design a dynamic adjustment mechanism based on the initial fertilizer application scheme set, including the classification of control levels, deviation triggering conditions, and model recalibration rules. Step S42: Obtain real-time growth status data of agricultural crops through real-time monitoring equipment, compare the real-time growth status data with the simulation data of the corresponding growth stage in the fertilizer regulation simulation dataset, calculate the deviation value of each growth indicator, and calculate the fertilizer regulation deviation parameter by combining the regulation weight of each indicator in the four-dimensional regulation matrix of agricultural crops. Step S43: Determine whether to trigger the dynamic adjustment mechanism based on the fertilizer regulation deviation parameter. If the deviation parameter is within the preset acceptable range, maintain the current fertilizer application plan. If the deviation parameter exceeds the acceptable range, match the corresponding regulation level according to the deviation size and start the fertilizer plan re-optimization process corresponding to the crop growth-fertilizer response model. Step S44: In the fertilizer scheme re-optimization process, with the goal of reducing fertilizer regulation deviation parameters, the fertilizer effect parameters and soil nutrient migration parameters in the crop growth-fertilizer response model are dynamically adjusted to regenerate the fertilizer regulation simulation dataset; based on the new fertilizer regulation simulation dataset and combined with real-time soil environmental data and meteorological forecast data, the optimal fertilizer regulation scheme for each growth stage is output. Step S45: Transform the optimal fertilizer regulation plan into control instructions that the fertilizer application execution system can recognize and push them to the field fertilization equipment simultaneously; at the same time, display the crop growth prediction comparison, soil nutrient change prediction, and fertilizer utilization efficiency change before and after the plan adjustment through a visualization platform, and generate a dynamic report on fertilizer regulation.

Images