Garlic planting whole-process database service system based on agricultural big data

Abstract

The invention provides a garlic planting full-process database service system based on agricultural big data, which relates to the technical field of agricultural planting and comprises a planting planning module, a seeding management module, a growth monitoring module, a topdressing irrigation management module, a pest control module and a harvesting decision module. The planting planning module accurately matches varieties and base fertilizer formulas according to soil and yield data; the seeding management and growth monitoring module dynamically optimizes planting measures in combination with real-time data and historical experience; the topdressing irrigation module, the pest control module and the harvesting decision-making module utilize technologies such as image recognition and spectrum detection to realize a precise and scientific management strategy, effectively improve the resource utilization rate, reduce the influence of environmental fluctuation on garlic growth, reduce the use of chemical agents, realize full-process data driving by integrating multi-source data, and improve the management efficiency. Meanwhile, a complete planting file is formed to support strategy iteration, large-scale and efficient development of garlic planting is facilitated, and then the economic benefits of garlic planting are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural planting technology, and more specifically, to a database service system for the entire garlic planting process based on agricultural big data. Background Technology

[0002] Against the backdrop of the rapid development of smart agriculture, garlic cultivation, as an important agricultural production field, still faces many limitations in existing technologies.

[0003] Currently, garlic cultivation management relies heavily on traditional experience. Although sensors are introduced to collect soil temperature, humidity, and meteorological data in some stages, the data is scattered and independent, forming "data silos" that cannot support scientific decision-making throughout the entire cultivation process. Key aspects such as variety selection, topdressing, and irrigation lack quantitative analysis models based on soil characteristics, target yield, and growth stage, leading to resource waste and yield loss. In the face of pest and disease risks and weather changes, manual inspection and experience-based judgment are insufficient to provide early warnings and dynamic responses. Overuse of chemical agents has also caused agricultural product safety and ecological environment problems, seriously restricting the large-scale and efficient development of garlic cultivation. Summary of the Invention

[0004] The purpose of this invention is to provide a database service system for the entire garlic planting process based on agricultural big data. This system can achieve precise decision-making and dynamic risk response through the deep integration of agricultural big data with the entire planting process, which is conducive to the large-scale and efficient development of garlic planting and thus improves the economic benefits of garlic planting.

[0005] In view of this, embodiments of this application provide a database service system for the entire garlic planting process based on agricultural big data, including:

[0006] The planting planning module obtains soil testing data and target yield for the target area, determines the garlic varieties suitable for the target area environment and the best sowing date based on the agricultural big data, and calculates the base fertilizer formula to generate sowing and fertilization plans.

[0007] The sowing management module sows according to the preset sowing plan and records sowing data, obtains soil temperature and humidity data, compares it with historical data in the agricultural big data database to evaluate seedling conditions, and adjusts the covering measures and soil moisture in the target area in conjunction with meteorological forecast data.

[0008] The growth monitoring module acquires soil nutrient data in real time to obtain environmental factors, and monitors plant growth and leaf color changes through image recognition technology to obtain growth indicators. It constructs a correlation model between environmental factors and growth indicators to predict each key growth stage.

[0009] The topdressing and irrigation management module implements a phased topdressing and irrigation strategy based on soil temperature and humidity data, soil nutrient data, and weather forecasts, and dynamically generates precise topdressing and irrigation plans according to the growth stage.

[0010] The pest and disease control module monitors signs of pest occurrence in the target area, and based on current environmental conditions and historical disease data from the agricultural big data database, it provides early warnings of pest and disease risk levels in order to generate pest and disease control plans that combine biological and chemical control.

[0011] The harvest decision module analyzes growth stage data, detects bulb maturity using near-infrared spectroscopy, and combines historical harvest times from an agricultural big data database with weather forecasts to determine the optimal harvest time window.

[0012] In some embodiments, determining the garlic variety suitable for the target area environment and the optimal planting date includes:

[0013] Obtain the average daily temperature data and last frost date data for the target planting area from the agricultural big data database for the past 10 years;

[0014] Obtain the biological zero point, effective accumulated temperature requirement, and accumulated temperature model of the target garlic variety, and determine the planting window period that meets the accumulated temperature requirement for more than 90% of the years;

[0015] By combining historical data on the last frost date, and avoiding the damage caused by late frost, the sowing date can be determined.

[0016] In some embodiments, soil temperature and humidity data are acquired and compared with historical data from an agricultural big data database to assess emergence conditions, including:

[0017] Temperature and humidity sensors are arranged in a grid pattern at preset intervals in the planting area;

[0018] The system automatically collects temperature and humidity data at preset intervals at depths of 5cm and 10cm in the soil.

[0019] The collected soil temperature and humidity data are compared with historical data from the same period in the agricultural big data database to calculate the temperature and humidity deviation index.

[0020] When the deviation index exceeds the preset threshold, soil moisture adjustment measures are generated to meet the conditions for seedling emergence.

[0021] In some embodiments, growth indicators are obtained by monitoring plant growth and leaf color changes using image recognition technology, including:

[0022] Acquire plant growth images of the target area using drone image acquisition equipment;

[0023] Using deep learning models to identify plant coverage and leaf color in images;

[0024] Calculate the Normalized Difference Vegetation Index (NDVI) to quantify plant growth status and assess plant health to form growth indicators.

[0025] Key growth stages include:

[0026] Screen relevant variables from environmental factors and perform data standardization;

[0027] By using the random forest algorithm, the corresponding environmental factors are associated with the growth index to obtain a multi-input multi-output association model;

[0028] The data is divided into training and validation sets in a 7:3 ratio. The model is trained by cross-validation, and the root mean square error (RMSE) is used as the loss function to obtain the final association model.

[0029] The key environmental factors for output are determined by ranking the importance weights of environmental factors in the association model, and the thresholds of key environmental factors for plant transformation at each stage are obtained through agricultural big data.

[0030] The latest key environmental factor data and growth index data are input into the trained correlation model. The model outputs the probability of entering the next growth stage in the future. When the probability is ≥70%, it is judged as a reliable prediction.

[0031] In some embodiments, based on soil temperature and humidity data, soil nutrient data, and weather forecasts, a phased topdressing and irrigation strategy is implemented, dynamically generating a precise topdressing and irrigation plan according to the growth stage, including:

[0032] Basic topdressing and irrigation plans for different growth stages of plants are obtained from agricultural big data.

[0033] Obtain the key time points of the current forecast and weather forecast, and dynamically adjust the time of the next topdressing irrigation according to the basic topdressing irrigation plan;

[0034] Based on soil temperature and humidity data, soil nutrient data, and weather forecasts, as well as the target soil moisture in the basic topdressing irrigation plan, the amount of topdressing irrigation for the next topdressing is dynamically adjusted to form a precise topdressing irrigation plan.

[0035] In some embodiments, monitoring signs of pest occurrence in target areas, and based on current environmental conditions and historical disease incidence data from an agricultural big data database, issuing early warnings of pest risk levels to generate pest control plans that combine biological and chemical control, including:

[0036] The target area is laid out in a grid, environmental data and plant growth images of the target area are obtained, image recognition algorithms are used to identify pest characteristics on leaves, and pest quantity and species distribution are obtained through pest trapping equipment.

[0037] Obtain pest and disease occurrence records for the target area from the agricultural big data database during the same period, and simultaneously retrieve environmental data for the corresponding period to establish a historical pest and disease-environment database.

[0038] By combining the biological characteristics and environmental data of major pests and diseases in the target area, and comparing them with historical pest and disease-environment databases, a risk assessment index system is established. The analytic hierarchy process (AHP) is used to determine the weight of each assessment index, and the pest and disease risk is quantitatively assessed to obtain a risk level score.

[0039] Based on the risk level score and historical records in the agricultural big data database, biological and chemical control plans are determined, and the control plans are dynamically optimized and adjusted in real time based on the control effect and subsequent monitoring data.

[0040] In some embodiments, growth stage data are analyzed, bulb maturity is detected using near-infrared spectroscopy, and the optimal harvest time window is determined by combining historical harvest times from an agricultural big data database with weather forecasts, including:

[0041] Near-infrared spectroscopy was used to obtain spectral data of bulbs, and a database of the correspondence between spectral data and maturity levels was established by combining the predicted key growth stages.

[0042] A regression model between spectral data and maturity was established using partial least squares (PLS) to calculate the predicted maturity value of the sample and determine whether the target area has entered the harvest preparation stage.

[0043] When the target area enters the harvest preparation stage, the harvest time data of the same variety of garlic in the target area in recent years are retrieved from the agricultural big data database. Combined with the meteorological data of the corresponding period, the optimal harvest time window is determined.

[0044] This invention provides a full-process database service system based on agricultural big data, achieving technological breakthroughs through the following methods:

[0045] Data-driven closed-loop process: Integrating multi-source data such as soil testing, meteorology, plant growth, and historical planting data, a full-cycle data association model is established from planting planning to harvest decision-making, breaking down data barriers between stages. For example, the variety and base fertilizer plans output by the planting planning module can serve as the basic parameters for the growth monitoring module to build an environment-growth index association model.

[0046] Precision decision support: Based on the quantitative relationship between soil nutrient baseline and target yield, the base fertilizer formula is calculated. Through image recognition technology, growth indicators such as leaf color and growth status are dynamically acquired to achieve precision in variety matching and growth stage prediction. Near-infrared spectroscopy technology is combined to detect bulb maturity, replacing traditional manual judgment and quantifying maturity into objective indicators such as starch content, thereby improving the scientific nature of harvest decisions.

[0047] Dynamic risk response mechanism: The sowing management module compares soil temperature and humidity with historical seedling data in real time, and dynamically adjusts covering measures in conjunction with weather forecasts to reduce the impact of environmental fluctuations on seedling rate; the pest and disease control module uses coupled analysis of environmental conditions and historical disease data to provide early warning of risk levels and generate "biological + chemical" combined control plans to reduce reliance on chemical agents.

[0048] Full-process traceability and optimization: The detection data, plan records, and execution results generated at each stage are all stored in the database to form a complete planting file, providing data accumulation for subsequent planting strategy optimization (such as iterating risk assessment models through historical pest and disease data).

[0049] By deeply integrating agricultural big data with the entire planting process, a paradigm shift from "experience-based planting" to "data-driven planting" can be achieved. This effectively solves problems such as low precision, delayed response, and fragmented management in existing technologies, providing a systematic solution for large-scale and efficient garlic planting. Precise decision-making and dynamic risk response are conducive to the large-scale and efficient development of garlic planting, thereby improving the economic benefits of garlic planting. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a module architecture diagram of a database service system provided in an embodiment of the present invention;

[0052] Figure 2 This is a flowchart illustrating how the planting planning module of this invention determines the variety and sowing date.

[0053] Figure 3 This is a flowchart illustrating how the growth monitoring module of this invention acquires soil temperature and humidity data to assess seedling emergence conditions.

[0054] Figure 4 This is a flowchart illustrating how the growth monitoring module acquires growth indicators according to an embodiment of the present invention.

[0055] Figure 5 This is a flowchart illustrating how the growth monitoring module of this invention predicts various key growth stages.

[0056] Figure 6 This is a flowchart illustrating the generation of a precise topdressing irrigation plan by the topdressing irrigation management module in an embodiment of the present invention.

[0057] Figure 7 This is a flowchart illustrating the generation of pest and disease control schemes by the pest and disease control module in an embodiment of the present invention.

[0058] Figure 8 This is a flowchart illustrating how the harvest decision module of this invention determines the optimal harvest time window. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0060] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0061] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0062] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0063] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0064] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0065] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other.

[0066] Traditional garlic cultivation relies heavily on experience-based management. Taking Jinxiang, Shandong Province as an example, past variety selection often resulted in yield reductions of 10%-15% due to neglecting accumulated temperature matching; indiscriminate irrigation led to water waste rates exceeding 30%; and lagging pest and disease control caused an average annual yield loss of 20%. The development of agricultural big data technology offers a new path to solving these problems. By integrating historical planting data, real-time environmental data, and crop growth data, scientific and precise planting decisions can be achieved. Currently, there is an urgent need to build an intelligent service system covering the entire planting cycle to solve key technical issues such as multi-source data fusion analysis, growth stage prediction, and dynamic control strategy generation.

[0067] Therefore, this application provides a database service system for the entire garlic planting process based on agricultural big data, referring to... Figure 1 ,include:

[0068] The planting planning module 100 acquires soil testing data and target yield for the target area, determines the garlic variety suitable for the target area environment and the optimal sowing date based on the agricultural big data database, and calculates the base fertilizer formula to generate sowing and fertilization plans.

[0069] Soil nutrient analyzers are used to sample the target area, and a standardized data interface is used to read the soil test report, which includes basic physicochemical indicators: pH value (accuracy ±0.1), organic matter content (g / kg, accuracy ±5%), EC value (mS / cm, accuracy ±2%), and nutrient parameters: available nitrogen (mg / kg), available phosphorus (mg / kg), and available potassium (mg / kg). At the same time, a verification engine automatically identifies abnormal values ​​(such as pH>9.0 or EC>4.0mS / cm triggering manual verification) to ensure that the soil test data integrity rate is ≥99%.

[0070] Among these, determining the garlic variety suitable for the target area's environment and the optimal planting date involves referring to... Figure 2 ,include:

[0071] S101: Obtain the average daily temperature data and last frost date data for the target planting area over the past 10 years from the agricultural big data database.

[0072] Connecting to the target region's climate database and the National Meteorological Information Center's CIMISS system, it stores nearly 30 years of daily meteorological data according to administrative divisions, including daily average temperature, accumulated temperature, last frost day, and frost-free period (resolution 1km×1km, such as the average temperature in 2023 being 14.8℃, and the average last frost day being March 22).

[0073] S102: Obtain the biological zero point, effective accumulated temperature requirement, and accumulated temperature model of the target garlic variety, and determine the planting window period that meets the accumulated temperature requirement for more than 90% of the years.

[0074] The data comes from an agricultural big data variety database, which includes variety data provided by eight research institutions, including the Institute of Vegetables and Flowers of the Chinese Academy of Agricultural Sciences. Each variety is accompanied by a phenological map of its growth period. The database includes biological parameters of more than 120 garlic varieties (such as the biological zero point of 5.8℃ and effective accumulated temperature of 1100℃·d for "Longsuan No. 3" of the Chinese Academy of Agricultural Sciences, and the biological zero point of 6.2℃ and effective accumulated temperature of 1150℃·d for the entire growth period for "Pizhou White Garlic") as well as 30 years of daily climate data (resolution 1km×1km).

[0075] Establish a variety-climate adaptation model:

[0076] Based on the average daily temperature over the past 10 years (75% of which is between 12-25℃), the fitness level was calculated using a Python script:

[0077] Fit = 0.4·f(T) + 0.3·f(K) + 0.2·f(Dfrost) + 0.1·f(frost-free period)

[0078] f(T) is the daily average temperature matching function (linear score in the 12-25℃ range), f(K) is the soil fertility matching function, f(Dfrost) is the frost day compliance rate, and f(frost-free period) is the effective accumulated temperature compliance rate.

[0079] Then, varieties with a suitability of ≥90% are automatically selected, and a recommendation list is generated. Based on the average daily temperature data of the past 10 years, the earliest / latest sowing date that meets the effective accumulated temperature requirement of the target variety is calculated, and the time interval with a confidence level of ≥90% is selected.

[0080] S103: Based on historical data on the last frost day, avoid the damage caused by late frost to determine the sowing date.

[0081] Using daily average temperature data from the past 10 years, data cleaning and outlier removal were performed using the NumPy library. The cumulative accumulated temperature was dynamically calculated using a sliding window algorithm. For example, for the "Jinxiang Purple Garlic" variety, the effective accumulated temperature threshold was set to ≥1200℃·d based on its biological characteristics. By constructing a probability distribution model through simulation, the safe sowing period in Jinxiang, Shandong was finally determined to be from September 25 to October 10, which covers 92% of the historical meteorological conditions.

[0082] The system integrates a frost warning API to monitor changes in the last frost date in real time. When the actual last frost date is delayed by more than 3 days compared to the historical average, an intelligent sowing date adjustment mechanism is triggered: based on historical accumulated temperature data and variety growth models, the system automatically generates the adjusted optimal sowing date and pushes warning information to growers via SMS and system pop-ups.

[0083] The sowing management module 200 sows according to the preset sowing plan and records the sowing data. It acquires soil temperature and humidity data, compares it with historical data in the agricultural big data database to evaluate the emergence conditions, and adjusts the covering measures and soil moisture in the target area in conjunction with meteorological forecast data.

[0084] This includes acquiring soil temperature and humidity data, comparing it with historical data from an agricultural big data database to assess seedling emergence conditions, and referring to... Figure 3 ,include:

[0085] S201: Temperature and humidity sensors are arranged in a grid pattern at preset intervals in the planting area;

[0086] After sowing, establish a soil moisture monitoring network:

[0087] Sensor deployment plan: (taking a 500-mu production area in Jinxiang, Shandong as an example)

[0088] A 15m × 15m square grid was used to deploy 222 monitoring nodes (including the boundary reinforcement zone), achieving 100% coverage of the land parcel. Each node is equipped with dual depth sensors.

[0089] 5cm depth: Monitoring surface soil temperature (affecting germination rate), using Decagon EC-5 sensor (accuracy ±0.25℃, response time ≤30 seconds);

[0090] 10cm depth: Monitor soil moisture in the root zone (affecting seedling uniformity), equipped with SoilQuest101 humidity sensor (accuracy ±2%RH, salt-alkali resistant design suitable for the Jinxiang Yellow River irrigation area);

[0091] For low-lying and flood-prone areas (approximately 15% of the total area), the grid is densified to 10m×10m, and conductivity sensors (EC-5 extension module, accuracy ±1%) are added to monitor salt migration caused by irrigation in real time (threshold setting: EC≤1.5mS / cm).

[0092] S202: Automatically collect temperature and humidity data at soil depths of 5cm and 10cm at preset intervals;

[0093] During normal periods, data is automatically collected every 12 hours. During the critical period of 7 days after sowing, the frequency is increased to once every 3 hours (dynamically adjusted through the edge computing unit). At the same time, the sensor is zero-point calibrated monthly using a standard temperature and humidity chamber (accuracy ±0.1℃ / ±1%RH) to ensure that the long-term monitoring error is ≤3%.

[0094] S203: Compare the collected soil temperature and humidity data with historical data from the same period in the agricultural big data database, and calculate the temperature and humidity deviation index;

[0095] Soil temperature and humidity data from the same period (0-15 days after sowing) over the past 10 years were extracted from the agricultural big data database to establish a variety-specific benchmark dataset (taking "Jinxiang Purple Garlic" as an example, historical average: 5cm soil temperature 12-16℃, 10cm humidity 60%-70%). Outliers were then removed, and missing data were filled in by linear interpolation to generate a benchmark curve (time resolution: 1 hour, spatial resolution: grid cell).

[0096] The comprehensive evaluation index, Temperature and Humidity Deviation Index (DI), is defined by the following formula:

[0097]

[0098] T cur / H cur Current measured temperature / humidity (average at 5cm / 10cm depth), T his / H his Historical average temperature / humidity values ​​for the same period (taken from the benchmark library), μ T / μ H Historical data standard deviation (reflecting regional climate stability, e.g., μ in Jinxiang area) T =1.5%, μ H =5%).

[0099] S204: When the deviation index exceeds the preset threshold, soil moisture adjustment measures are generated to meet the conditions for seedling emergence.

[0100] Threshold setting:

[0101] Slight deviation: DI = 0.1-0.2 (Yellow alert, initiate field inspection)

[0102] Severe deviation: DI≥0.2 (Red alert, automatic soil moisture adjustment triggered)

[0103] For example, the temperature and humidity data detected at a certain time are as follows:

[0104] 5cm soil temperature 11℃ (historical average 13℃) cur -T his =-2℃))

[0105] Humidity at 10cm: 55% (historical average: 65%) cur -H his =-10%)),

[0106] calculate This will trigger a severe deviation warning.

[0107] Based on relevant technologies provided by agricultural big data, we have developed intelligent soil moisture regulation technologies, including temperature control strategies and water control strategies.

[0108] Temperature control strategies include:

[0109] Mulch film mulching strategy: When the soil temperature at 5cm depth is <12℃ and DI≥0.2, mechanized mulch film mulching should be initiated (using 0.01mm thick polyethylene mulch film with 90% light transmittance). The soil temperature can be increased by 2-3℃ within 48 hours after mulching (actual data from 2023: 10℃ before mulching → 13.5℃ after mulching).

[0110] Frost emergency strategy: Combine with AccuWeather frost warning (for the next 48 hours)

[0111] With a minimum temperature ≤0℃ probability >30%, it automatically adds straw mulch (5-8cm thick) to increase the ground temperature by 1-2℃ and avoid damage from late frost (2018 application case: pre-frost mulch reduced the seedling frost damage rate from 40% to 8%).

[0112] Water regulation strategies include:

[0113] Dynamic calculation strategy for drip irrigation volume: When the humidity at 10cm depth is <60% and DI ≥ 0.2, the drip irrigation system is activated. The formula for calculating the volume of water irrigated per irrigation is as follows:

[0114]

[0115] H target Target humidity (e.g., the optimal germination range for "Jinxiang purple garlic" is set at 65%);

[0116] ρ: Soil bulk density; A: Irrigated area (hectares); η: Irrigation efficiency (90%, including pipeline losses)

[0117] For example, when the humidity of a 50-hectare plot is 55%,

[0118] This is equivalent to approximately 10.8 mm of precipitation.

[0119] Water quality optimization strategies:

[0120] A water quality sensor is installed to monitor the pH value of irrigation water in real time (target 6.0-7.5). When the detected value is <6.0, calcium hydroxide is automatically added (addition amount = 0.5kg / 1000m³). 3 When the concentration of water is >7.5, add citric acid (addition amount = 0.3 kg / 1000 m³). 3 Water) to ensure that the pH of the root zone soil is stable at 6.5±0.3.

[0121] The growth monitoring module 300 acquires soil nutrient data in real time to obtain environmental factors, and uses image recognition technology to monitor plant growth and leaf color changes to obtain growth indicators. It then constructs a correlation model between environmental factors and growth indicators to predict each key growth stage.

[0122] The soil moisture monitoring network also includes a soil nutrient dynamics monitoring network. When constructing the sensor network, following the same grid strategy, multi-parameter soil detectors are synchronously installed and buried at the center of each 15m × 15m grid for real-time monitoring.

[0123] Nitrogen forms: Concentration (accuracy ±5%, response time ≤2 minutes), the measured average value during the bolting stage in 2024 was 48 mg / L (the nitrogen fertilizer warning threshold can be set to 50 mg / L);

[0124] Phosphorus and potassium availability: K + Concentration (accuracy ±3%), combined with zone characteristics (the early warning threshold for readily available phosphorus can be set to 20 mg / L);

[0125] pH and conductivity: accuracy ±0.1 / ±1%, real-time feedback on the impact of irrigation water quality on soil chemical properties (e.g., pH value fluctuates from 7.2 to 6.8 after drip irrigation).

[0126] The data is collected and uploaded every 12 hours during the normal period, and every 3 hours during the critical growth period (bulb enlargement period).

[0127] In the growth monitoring module 300, growth indicators are obtained by monitoring plant growth and leaf color changes through image recognition technology, with reference to... Figure 4 ,include:

[0128] S301: Acquire plant growth images of a target area using drone image acquisition equipment;

[0129] Each week, during periods of uniform sunlight, a drone equipped with a multispectral camera is used to take aerial photos of the target area. The aerial photography altitude can be 100m, with a ground resolution of 0.1m (flight speed 5m / s, heading overlap rate 80%), in order to acquire images of all plant growth in the target area.

[0130] S302: Use deep learning models to identify plant coverage and leaf color in images;

[0131] First, the image data is preprocessed. Whiteboard calibration data (standard whiteboard with known reflectivity) is used to correct sensor response deviations to ensure NDVI calculation accuracy. Then, image stitching is performed based on RTK differential positioning data (accuracy ±2cm) to generate a seamless mosaic. Then, the dark pixel method is used to remove aerosol effects, such as the average summer aerosol optical thickness (AOD) of 0.3 in Jinxiang area, to reduce reflectivity error.

[0132] The parsing of preprocessed images using deep learning models includes:

[0133] U-Net (backbone network ResNet50) was trained on the image dataset of the target area to parse the images to obtain the plant coverage. The RGB images were converted to the HSV color space, and the saturation (S component) was extracted as the leaf color index to determine the leaf color.

[0134] The Faster R-CNN+ResNeXt101 architecture was used, trained on a leaf blight dataset (2000+ labeled lesions, mAP=0.88), to identify lesions ≥2mm in diameter. 2 lesions.

[0135] S303: Calculate the Normalized Difference Vegetation Index (NDVI) to quantify plant growth status and assess plant health to form growth indicators.

[0136] NDVI (Normalized Difference Vegetation Index) is an important indicator reflecting vegetation growth status, and its calculation formula is as follows:

[0137]

[0138] Among them, R 842 R represents the near-infrared reflectance at a wavelength of 842 nm. 668 The reflectance is measured in the red light band at a wavelength of 668nm. Vegetation has high reflectance in the near-infrared band (842nm), but low reflectance in the red light band (668nm) due to chlorophyll absorption. By comparing the difference between near-infrared and red light reflectance and the ratio of their sum, the influence of background noise such as terrain and lighting can be eliminated, highlighting vegetation information.

[0139] NDVI values ​​typically range from -1 to 1. Negative values ​​indicate non-vegetated areas (such as water bodies or bare land); values ​​near 0 indicate extremely low vegetation cover; and positive values ​​indicate higher vegetation cover and better growth conditions (such as forests or dense farmland). The current plant status is determined based on the NDVI value.

[0140] In the growth monitoring module 300, a correlation model between environmental factors and growth indicators is constructed to predict each key growth stage, referring to... Figure 5 ,include:

[0141] S311: Screen relevant variables among environmental factors and perform data standardization;

[0142] Environmental factors with |ρ|≥0.5 were screened using Spearman rank correlation coefficient. Core parameters were retained from 20+ candidate variables. Principal component analysis was performed on the retained environmental factors. The cumulative variance contribution rate of the first 3 principal components was 85%, confirming that there were no redundant variables (e.g., the correlation between soil pH and electrical conductivity was >0.9, so only pH was retained).

[0143] Z-score standardization of environmental factors:

[0144]

[0145] Where μ is the average value of the target region over the past 5 years, σ is the standard deviation (e.g., a sample measured 42 mg / L → after standardization -1.0), and then the data are aligned according to the garlic growth degree day (GDD), the sowing day is set to GDD = 0, and the effective accumulated temperature of ≥6℃ is accumulated daily to ensure that the data of different years are aligned in time and space.

[0146] S312: By using the random forest algorithm, the corresponding environmental factors are associated with the growth index to obtain a multi-input multi-output association model;

[0147] Constructing the algorithm architecture, environmental factors (daily average temperature, soil) The model uses PAR (particulate matter ratio), atmospheric humidity, and 5cm soil temperature as input layers, and stage transition probabilities (seedling stage → bolting stage, bolting stage → bulb enlargement stage) as output layers. Model initialization is achieved by configuring model parameters in the scikit-learn framework.

[0148] S313: Divide the data into training and validation sets in a 7:3 ratio, train the model through cross-validation, and use the root mean square error (RMSE) as the loss function to obtain the final association model.

[0149] The training and validation sets are divided in a 7:3 ratio, such as 15,000 samples from 2019 to 2023 and 6,000 samples from 2024, to ensure a balanced distribution of samples at each growth stage (e.g., 25% of samples are from the bolting stage). Then, 5-fold cross-validation is implemented, maintaining the integrity of the time series during each fold of training, and the root mean square error (RMSE) is used as the loss function to calculate the deviation between the predicted number of days and the actual number of days.

[0150] A time window is constructed by combining environmental factors with growth indicators that lag by 1-3 days. Training is terminated when the RMSE of the validation set increases by more than 5% for 10 consecutive rounds to prevent overfitting and ultimately obtain the correlation model.

[0151] S314: Determine the key environmental factors for output based on the importance weight ranking of environmental factors in the association model, and obtain the threshold of key environmental factors for plant transformation at each stage through agricultural big data.

[0152] The contribution of each environmental factor to node purity is calculated, the top three key environmental factors are selected, and the average environmental factor value of the stage transition date in the past 10 years is extracted from the agricultural big data as the threshold of key environmental factors.

[0153] S315: Input the latest key environmental factor data and growth index data into the trained correlation model. The model outputs the probability of entering the next growth stage in the future. When the probability is ≥70%, it is judged as a reliable prediction.

[0154] The system receives real-time sensor data from the target area, removes noise using Kalman filtering, and then maps the real-time data into the model input format.

[0155] After inputting it into the model, the model will simultaneously output the probability of entering the next stage in the next 7 days. For example, the probability of entering the next stage within 7 days is 0.75 for bolting stage → swelling stage, and 0.40 for swelling stage → maturity stage, which is 40% probability of entering the next stage within 7 days. When the probability is ≥70% for 2 consecutive days and at least one key factor exceeds the threshold (such as accumulated temperature exceeding 600℃·d), a stage transition warning is triggered.

[0156] Then the prediction results are verified. For example, the input data for April 5th is: average daily temperature 18℃ (exceeding the threshold of 15℃). The concentration was 48 mg / L (≥45), the model output probability was 72%, the actual bolting date was April 8th, the prediction was 3 days in advance, which met the threshold rule. If the predicted probability is >70% for 3 consecutive days but does not actually change, the model will be automatically fine-tuned (the learning rate will be reduced by 0.1 and the most recent 100 samples will be retrained).

[0157] The topdressing and irrigation management module 400, based on soil temperature and humidity data, soil nutrient data, and weather forecasts, implements a phased topdressing and irrigation strategy, dynamically generating precise topdressing and irrigation plans according to the growth stage, and referring to... Figure 6 ,include:

[0158] S401: Obtain basic topdressing and irrigation plans for different growth stages of plants from agricultural big data;

[0159] Based on the biological characteristics of garlic, the entire growth period is divided into five key stages: seedling stage, flower bud differentiation stage, bolting stage, bulb enlargement stage, and maturity stage. A basic plan matching the target plot's soil type, fertility, and target yield is retrieved from the agricultural big data database. This plan includes: target soil moisture for each stage (e.g., maintaining 65-70% field capacity during bulb enlargement) and recommended irrigation amounts (e.g., 15-20 m³ / h). 3 The formula for topdressing (e.g., N:P:K = 3:1:2) is determined by the application rate per mu per application. The feasibility of the plan is verified by combining historical meteorological data (e.g., increasing irrigation by 20% in drought years). The plan is then revised using the expert knowledge base in the agricultural big data database (e.g., increasing potassium fertilizer by 10% during the bulb enlargement period in the Huang-Huai-Hai Plain region).

[0160] S402: Obtain the key time points of the current forecast and weather forecast, and dynamically adjust the time of the next topdressing irrigation according to the basic topdressing irrigation plan;

[0161] Based on the data provided by the sowing management module 200 and the growth monitoring module 300, including the predicted time of key growth stages, weather forecast data, and the covering measures and soil moisture of the target area, the execution time in the basic plan is adjusted to generate a revised topdressing and irrigation schedule.

[0162] S403: Based on soil temperature and humidity data, soil nutrient data, and weather forecasts, as well as the target soil moisture in the basic topdressing irrigation plan, dynamically adjust the amount of the next topdressing irrigation to form a precision topdressing irrigation plan.

[0163] Combining soil temperature and humidity data, soil nutrient data, and weather forecasts obtained from the soil moisture monitoring network constructed in the sowing management module 200, based on the principle of nutrient balance:

[0164]

[0165] The basic fertilization amount is the recommended value for each stage in the basic plan, and the measured nutrient content is calculated by combining the results of real-time monitoring by soil sensors with irrigation and topdressing amounts to generate a precise plan.

[0166] Simultaneously, the sowing management module 200 and the growth monitoring module 300 are combined to monitor the plant status and soil moisture in real time, and compare them with the expected growth status in the agricultural big data database to adjust the topdressing and irrigation plan in real time.

[0167] The pest and disease control module 500 monitors signs of pest occurrence in the target area. Based on current environmental conditions and historical disease data from the agricultural big data database, it issues early warnings of pest and disease risk levels to generate pest and disease control plans that combine biological and chemical control methods. Figure 7 ,include:

[0168] S501: The target area is laid out in a grid, environmental data and plant growth images of the target area are obtained, the image recognition algorithm is used to identify pest characteristics on the leaves, and the distribution of pest quantity and species is obtained through pest trapping equipment.

[0169] By acquiring plant growth images from the growth monitoring module 300 and analyzing the results of these images, pest characteristics on the leaves are obtained, and the current environmental data and pest and disease status of the plant are determined. At the same time, the distribution of pest numbers and species is obtained through pest trapping equipment and recorded periodically.

[0170] S502: Obtain the pest and disease occurrence records of the target area in the same period from the agricultural big data database, and at the same time retrieve the environmental data of the corresponding period to establish a historical pest and disease-environment database.

[0171] Nearly 10 years of pest and disease records were obtained from an agricultural big data database and cataloged according to "year + pest / disease type + occurrence date + environmental data". Then, the data was linked with environmental data such as temperature, humidity, and soil nutrients of the corresponding year to build a "pest / disease-environment" database.

[0172] S503: Combining the biological characteristics and environmental data of major pests and diseases in the target area, a risk assessment index system is established by comparing with historical pest and disease-environment databases. The weight of each assessment index is determined by the analytic hierarchy process (AHP), and the risk of pests and diseases is quantitatively assessed to obtain a risk level score.

[0173] Based on the biological characteristics of major pests and diseases in the target area (such as aphids and leaf blight) (e.g., reproductive temperature threshold, spread humidity conditions), key indicators are screened from environmental factors and pest characteristics to form an initial indicator set (e.g., temperature, humidity, pest density, lesion coverage, historical occurrence frequency, etc.), and a three-layer indicator system is constructed:

[0174] Target layer: Pest and disease risk level;

[0175] Criterion layer: environmental suitability, pest occurrence intensity, historical risk factors;

[0176] Indicator layer: Specific quantifiable indicators (such as temperature suitability, number of pests trapped, and number of occurrences in the same period in history).

[0177] Construct a judgment matrix for pairwise comparisons between indicators, and calculate the weight of each indicator (must pass a consistency test, CI < 0.1).

[0178] S504: Based on the risk level score and historical records in the agricultural big data database, determine the biological and chemical control plans, and dynamically optimize and adjust the control plans in real time based on the pest and disease control effects and subsequent monitoring data.

[0179] The environmental data and pest characteristic data collected in real time are standardized by using extreme value standardization or Z-score standardization to convert indicators of different dimensions into score values ​​in the 0-1 range.

[0180] Based on the weights determined by the analytic hierarchy process (AHP), the standardized indicators are weighted and summed to obtain the risk level score (0-100 points) of the grid cell. The formula is as follows:

[0181]

[0182] Where w i For the indicator weights, s i Standardized score for a single indicator.

[0183] Define risk level ranges (e.g., low risk 0-50 points, medium risk 50-80 points, high risk >80 points), and generate a grid-level risk distribution map by combining the pest and disease control thresholds.

[0184] Based on historical prevention and control records in the agricultural big data database, a prevention and control program database corresponding to different risk levels is established, including: biological control measures (release of natural enemies, types and dosages of biological pesticides); chemical control measures (pesticide varieties, application dosages, and intervals); and physical control measures (density of trapping equipment and color board layout). Based on real-time risk scores, the corresponding prevention and control program is automatically matched, with priority given to recommending environmentally friendly measures (such as using only biological control in low-risk areas) and using combined biological and chemical control in high-risk areas.

[0185] After the implementation of prevention and control measures, data on the effectiveness of prevention and control (pest reduction rate, lesion control rate) and environmental data are continuously collected to calculate the deviation between the actual effect and the expected effect.

[0186] If the deviation exceeds the preset threshold (e.g., pest reduction rate <60%), the scheme optimization is triggered, and the control measures parameters are adjusted (e.g., increase pesticide dosage by 10% or shorten the application interval by 2 days); the optimized scheme and effect data are fed back to the historical database to update the control scheme database.

[0187] The harvest decision module 600 analyzes growth stage data, detects bulb maturity using near-infrared spectroscopy, and combines historical harvest times from an agricultural big data database with weather forecasts to determine the optimal harvest time window. Figure 8 ,include:

[0188] S601: Obtain spectral data of bulbs through near-infrared spectroscopy, and establish a database of the correspondence between spectral data and maturity level by combining the predicted key growth stages.

[0189] Using a near-infrared spectrometer, several sampling points were randomly selected in the target area in a grid pattern. Spectral data of several garlic bulbs were collected at each point (avoiding areas with epidermal damage). The collection time was fixed from 10:00 AM to 12:00 PM (to avoid the influence of dew). Three spectra were collected for each bulb (top, middle, and bottom), and the average spectral value was taken.

[0190] Savitzky-Golay filtering was used to remove noise, combined with multivariate scattering correction (MSC) to eliminate the influence of particle size, and characteristic bands (such as 1180nm corresponding to water content and 1450nm corresponding to soluble solids) were extracted to compress the original spectrum to 20 characteristic variables.

[0191] S602: Use partial least squares (PLS) to establish a regression model between spectral data and maturity, calculate the predicted maturity value of the sample, and determine whether the target area has entered the harvest preparation stage;

[0192] The maturity of garlic is divided into three levels: immature, mature, and overripe. The preprocessed spectral data is correlated with the output data of the growth stage prediction model (such as the start time of bulb enlargement) to construct time series data on maturity evolution and establish a database model of "spectral characteristics-maturity index".

[0193] The database model is trained by randomly selecting 80% of the samples (approximately 1200 sets of spectral-index data) from the database as the training set and 20% as the validation set. Five-fold cross-validation is performed on the training set to prevent overfitting, thus forming the PLS model.

[0194] Real-time spectral data is acquired, preprocessed, and input into the PLS model. The model outputs a predicted maturity value (0-1). Combined with the grading standard, it is determined whether the model has entered the harvest period. When the maturity value of ≥70% of the sampling points is ≥0.8 for 3 consecutive days, and the prediction model shows that the bulb enlargement period has ended for ≥15 days, the model is determined to enter the harvest preparation stage.

[0195] S603: When the target area enters the harvest preparation stage, retrieve the harvest time data of the same variety of garlic in the target area in recent years from the agricultural big data database, and combine it with the meteorological data of the corresponding period to determine the optimal harvest time window.

[0196] The actual harvest dates of the same garlic variety in the target area over the past 5 years were retrieved from the agricultural big data database, and meteorological data (such as the number of consecutive sunny days, diurnal temperature range, and rainfall) for the corresponding period were matched to establish a meteorological-harvest time correlation model.

[0197] Obtain detailed 14-day forecast data from local meteorological departments (resolution ≤ 1km). 2 ), with particular attention to:

[0198] Precipitation probability: Avoid areas with moderate to heavy rain (precipitation probability ≥ 30%) within the next 7 days;

[0199] Temperature fluctuations: Select a daily average temperature range of 22-25℃ (which is beneficial for bulb dehydration and storage);

[0200] Wind speed conditions: Prioritize time periods with wind speeds ≤ level 4 (to reduce mechanical harvesting losses).

[0201] Based on the start date of the optimal harvest period predicted by the PLS model, and combined with historical data, the theoretical harvest window (±5 days) is determined and adjusted according to meteorological data: if there is precipitation within the forecast window, the window is extended to ≥3 consecutive sunny days after the rain; if the temperature remains >28℃, harvesting is started 2 days earlier (to avoid bulb aging). The optimal harvest time window is finally generated.

[0202] This invention constructs a full-process database service system based on agricultural big data, achieving technological breakthroughs through the following methods:

[0203] Data-driven closed-loop process: Integrating multi-source data such as soil testing, meteorology, plant growth, and historical planting data, a full-cycle data association model is established from planting planning to harvest decision-making, breaking down data barriers between stages. For example, the variety and base fertilizer plans output by the planting planning module can serve as the basic parameters for the growth monitoring module to build an environment-growth index association model.

[0204] Precision decision support: Based on the quantitative relationship between soil nutrient baseline and target yield, the base fertilizer formula is calculated. Through image recognition technology, growth indicators such as leaf color and growth status are dynamically acquired to achieve precision in variety matching and growth stage prediction. Near-infrared spectroscopy technology is combined to detect bulb maturity, replacing traditional manual judgment and quantifying maturity into objective indicators such as starch content, thereby improving the scientific nature of harvest decisions.

[0205] Dynamic risk response mechanism: The sowing management module compares soil temperature and humidity with historical seedling data in real time, and dynamically adjusts covering measures in conjunction with weather forecasts to reduce the impact of environmental fluctuations on seedling rate; the pest and disease control module uses coupled analysis of environmental conditions and historical disease data to provide early warning of risk levels and generate "biological + chemical" combined control plans to reduce reliance on chemical agents.

[0206] Full-process traceability and optimization: The detection data, plan records, and execution results generated at each stage are all stored in the database to form a complete planting file, providing data accumulation for subsequent planting strategy optimization (such as iterating risk assessment models through historical pest and disease data).

[0207] By deeply integrating agricultural big data with the entire planting process, a paradigm shift from "experience-based planting" to "data-driven planting" can be achieved. This effectively solves problems such as low precision, slow response, and fragmented management in existing technologies, providing a systematic solution for large-scale and efficient garlic planting. This is conducive to the large-scale and efficient development of garlic planting, thereby effectively improving the economic benefits of garlic planting.

[0208] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0209] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A database service system for the entire garlic planting process based on agricultural big data, characterized in that, include: The planting planning module obtains soil testing data and target yield for the target area, determines the garlic varieties suitable for the target area environment and the best sowing date based on the agricultural big data, and calculates the base fertilizer formula to generate sowing and fertilization plans. The sowing management module sows according to the preset sowing plan and records sowing data, obtains soil temperature and humidity data, compares it with historical data in the agricultural big data database to evaluate seedling conditions, and adjusts the covering measures and soil moisture in the target area in conjunction with meteorological forecast data. The growth monitoring module acquires soil nutrient data in real time to obtain environmental factors, and monitors plant growth and leaf color changes through image recognition technology to obtain growth indicators. It constructs a correlation model between environmental factors and growth indicators to predict each key growth stage. The topdressing and irrigation management module implements a phased topdressing and irrigation strategy based on soil temperature and humidity data, soil nutrient data, and weather forecasts, and dynamically generates precise topdressing and irrigation plans according to the growth stage. The pest and disease control module monitors signs of pest occurrence in the target area, and based on current environmental conditions and historical disease data from the agricultural big data database, it provides early warnings of pest and disease risk levels in order to generate pest and disease control plans that combine biological and chemical control. The harvest decision module analyzes growth stage data, detects bulb maturity using near-infrared spectroscopy, and combines historical harvest times from an agricultural big data database with weather forecasts to determine the optimal harvest time window.

2. The service system according to claim 1, characterized in that, Determine the garlic variety suitable for the target area environment and the optimal planting date, including: Obtain the average daily temperature data and last frost date data for the target planting area from the agricultural big data database for the past 10 years; Obtain the biological zero point, effective accumulated temperature requirement, and accumulated temperature model of the target garlic variety, and determine the planting window period that meets the accumulated temperature requirement for more than 90% of the years; By combining historical data on the last frost date, and avoiding the damage caused by late frost, the sowing date can be determined.

3. The service system according to claim 1, characterized in that, Acquire soil temperature and humidity data and compare them with historical data from agricultural big data databases to assess seedling emergence conditions, including: Temperature and humidity sensors are arranged in a grid pattern at preset intervals in the planting area; The system automatically collects temperature and humidity data at preset intervals at depths of 5cm and 10cm in the soil. The collected soil temperature and humidity data are compared with historical data from the same period in the agricultural big data database to calculate the temperature and humidity deviation index. When the deviation index exceeds the preset threshold, soil moisture adjustment measures are generated to meet the conditions for seedling emergence.

4. The service system according to claim 1, characterized in that, Growth indicators, including those obtained by monitoring plant growth and leaf color changes using image recognition technology, include: Acquire plant growth images of the target area using drone image acquisition equipment; Using deep learning models to identify plant coverage and leaf color in images; Calculate the Normalized Difference Vegetation Index (NDVI) to quantify plant growth status and assess plant health to form growth indicators.

5. The service system according to claim 1, characterized in that, Construct a correlation model between environmental factors and growth indicators to predict key growth stages, including: Screen relevant variables from environmental factors and perform data standardization; By using the random forest algorithm, the corresponding environmental factors are associated with the growth index to obtain a multi-input multi-output association model; The data was divided into training and validation sets in a 7:3 ratio. The model was trained by cross-validation, and the root mean square error (RMSE) was used as the loss function to obtain the final association model. The key environmental factors for output are determined by ranking the importance weights of environmental factors in the association model, and the thresholds of key environmental factors for plant transformation at each stage are obtained through agricultural big data. The latest key environmental factor data and growth index data are input into the trained correlation model. The model outputs the probability of entering the next growth stage in the future. When the probability is ≥70%, it is judged as a reliable prediction.

6. The service system according to claim 1, characterized in that, Based on soil temperature and humidity data, soil nutrient data, and weather forecasts, a phased topdressing and irrigation strategy is implemented. Precision topdressing and irrigation plans are dynamically generated according to the growth stage, including: Basic topdressing and irrigation plans for different growth stages of plants are obtained from agricultural big data. Obtain the key time points of the current forecast and weather forecast, and dynamically adjust the time of the next topdressing irrigation according to the basic topdressing irrigation plan; Based on soil temperature and humidity data, soil nutrient data, and weather forecasts, as well as the target soil moisture in the basic topdressing irrigation plan, the amount of topdressing irrigation for the next topdressing is dynamically adjusted to form a precise topdressing irrigation plan.

7. The service system according to claim 1, characterized in that, Monitor signs of pest occurrence in target areas, and based on current environmental conditions and historical disease data from agricultural big data databases, issue early warnings on pest risk levels to generate pest control plans that combine biological and chemical control methods, including: The target area is laid out in a grid, environmental data and plant growth images of the target area are obtained, image recognition algorithms are used to identify pest characteristics on leaves, and pest quantity and species distribution are obtained through pest trapping equipment. Obtain pest and disease occurrence records for the target area from the agricultural big data database during the same period, and simultaneously retrieve environmental data for the corresponding period to establish a historical pest and disease-environment database. By combining the biological characteristics and environmental data of major pests and diseases in the target area, and comparing them with historical pest and disease-environment databases, a risk assessment index system is established. The analytic hierarchy process (AHP) is used to determine the weight of each assessment index, and the pest and disease risk is quantitatively assessed to obtain a risk level score. Based on the risk level score and historical records in the agricultural big data database, biological and chemical control plans are determined, and the control plans are dynamically optimized and adjusted in real time based on the control effect and subsequent monitoring data.

8. The service system according to claim 1, characterized in that, Analyzing growth stage data, near-infrared spectroscopy was used to detect bulb maturity. Combined with historical harvest times from an agricultural big data database and weather forecasts, the optimal harvest time window was determined, including: Near-infrared spectroscopy was used to obtain spectral data of bulbs, and a database of the correspondence between spectral data and maturity levels was established by combining the predicted key growth stages. A regression model between spectral data and maturity was established using partial least squares (PLS) to calculate the predicted maturity value of the sample and determine whether the target area has entered the harvest preparation stage. When the target area enters the harvest preparation stage, the harvest time data of the same variety of garlic in the target area in recent years are retrieved from the agricultural big data database. Combined with the meteorological data of the corresponding period, the optimal harvest time window is determined.

Images