Iot-based intelligent agricultural management system and method

Abstract

The application relates to the technical field of agricultural management, in particular to a smart agricultural management system and method based on the Internet of Things. The method comprises the following steps: acquiring farmland geographic data and farmland environment images; extracting scene feature points of the farmland environment images, combining the farmland geographic data to perform farmland three-dimensional modeling, and generating a digital twin farmland; acquiring to-be-seeded seed data; analyzing development characteristics of the to-be-seeded seed data to obtain to-be-seeded seed development characteristic data; importing the to-be-seeded seed development characteristic data into the digital twin farmland to perform seeding simulation and growth simulation, and generating seeding simulation data and growth simulation data; and performing seeding profile analysis on the seeding simulation data to obtain a seeding profile graph. Through the combination of digital twinning, the Internet of Things and intelligent simulation technology, the precision and efficiency of agricultural seeding management are improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural management technology, and in particular to a smart agricultural management system and method based on the Internet of Things. Background Technology

[0002] Initially, agricultural production relied primarily on manual labor, with relatively traditional monitoring and management methods and a low level of informatization. With the development of information technology, the agricultural sector gradually introduced technologies such as sensors and automated equipment to improve production efficiency and accuracy. However, these early technologies were mostly limited to single devices and data collection, failing to achieve real-time interconnection and comprehensive utilization of information. The Internet of Things (IoT) technology enables various devices and sensors in the agricultural production process to connect via the internet, collecting data such as soil moisture, temperature, and climate in real time, and analyzing and processing this data through cloud platforms. Agricultural production can not only be managed with precision but also automated, such as automatic control of irrigation systems and intelligent fertilization, thereby greatly improving production efficiency and resource utilization. However, current traditional sowing methods are often based on experience, lacking targeted analysis of the developmental characteristics of different seeds, and cannot provide real-time feedback on the impact of sowing depth or location on crop growth, especially the imbalance of root-to-shoot ratio, leading to low precision and efficiency in agricultural sowing management. Summary of the Invention

[0003] Therefore, it is necessary to provide an IoT-based smart agriculture management system and method to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a smart agriculture management method based on the Internet of Things (IoT) is proposed, the method comprising the following steps:

[0005] Step S1: Acquire farmland geographic data and farmland environment images; extract scene feature points from farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland;

[0006] Step S2: Obtain seed data to be sown; analyze the developmental characteristics of the seed data to obtain seed developmental characteristic data; import the seed developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data;

[0007] Step S3: Perform sowing profile analysis on the sowing simulation data to obtain a sowing profile diagram; analyze the root-to-shoot ratio imbalance of the sowing profile diagram based on the growth simulation data, and adjust the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data;

[0008] Step S4: Use the Internet of Things to import the seeding optimization simulation data into the smart seeder for seeding operation and collect seeding operation data simultaneously; perform real-time seeding position correction on the seeding operation data to execute smart agricultural seeding management operations.

[0009] This invention, through the combination of 3D modeling and digital twin farmland, can accurately simulate the sowing and crop growth process in a virtual environment. Real-time data updates and analysis allow for fine-tuning based on environmental changes and farmland conditions. This enhanced precision ensures crops grow in the most suitable environment, maximizing yield. By acquiring and analyzing the developmental characteristics of seeds to be sown, customized sowing plans can be provided for different crop seeds, optimizing growth conditions for different crops. This personalized management improves germination rates and growth efficiency, reducing seed waste. Optimizing the root-to-shoot ratio is a key factor affecting plant health and yield. By analyzing imbalances in the root-to-shoot ratio, parameters such as sowing depth and soil fertilization can be adjusted promptly to avoid growth obstacles caused by imbalances, improving root development and leaf growth, and promoting overall healthy growth. Continuous collection and analysis of sowing simulation data, growth simulation data, and sowing operation data provide real-time decision support for agricultural managers. Specifically, growth simulation data can be used to predict crop growth trends and the likelihood of pests and diseases, allowing for proactive intervention to improve yield and quality. The introduction of intelligent seeders not only achieves precise sowing but also allows for dynamic adjustments based on real-time data. This automated operation significantly reduces human intervention and operational errors, improves sowing efficiency, and ensures crops grow at appropriate depths and spacing, contributing to even crop distribution and reducing waste and resource waste. Therefore, this invention improves the accuracy and efficiency of agricultural sowing management by combining digital twin, Internet of Things, and intelligent simulation technologies.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: Obtain farmland geographic data using GIS technology; acquire farmland environmental images using cameras;

[0012] Step S12: Standardize the coordinates of the farmland geographic data to generate standardized farmland geographic data;

[0013] Step S13: Perform feature point detection and matching on the farmland environment image, and extract scene feature points from the farmland environment image to obtain scene feature points;

[0014] Step S14: Combine standardized farmland geographic data and scene feature points to reconstruct farmland point clouds and generate initial farmland point cloud data;

[0015] Step S15: Perform 3D surface modeling on the initial point cloud data of the farmland to generate a digital twin farmland.

[0016] This invention acquires farmland geographic data using GIS technology and combines it with farmland environment images captured by cameras to comprehensively collect spatial information and environmental features of farmland, providing a high-quality data foundation for subsequent modeling and analysis. The application of GIS technology ensures the accuracy of the farmland geographic data, while the introduction of image data supplements it with rich information about the farmland environment. By standardizing the coordinates of the farmland geographic data, the consistency of data from different sources under the same coordinate system is ensured, eliminating errors caused by differences in coordinate systems. This step lays a solid foundation for subsequent reconstruction and modeling, guaranteeing high accuracy in 3D farmland modeling. Through feature point detection and matching of farmland environment images, key scene feature points are extracted, accurately capturing the geometric features of the farmland environment. These feature points provide necessary data for point cloud reconstruction of the farmland, further enhancing the accuracy and stability of the modeling process. Combining standardized farmland geographic data and extracted scene feature points for farmland point cloud reconstruction generates preliminary 3D point cloud data of the farmland. The point cloud data provides detailed structural information of the farmland space, providing important 3D spatial data support for further modeling and analysis. By performing 3D surface modeling on the initial point cloud data of farmland, digital twin farmland can be generated, accurately restoring the 3D terrain and environment of the farmland. This modeling method not only improves the visualization effect of farmland data, but also provides virtual environment support for subsequent intelligent management, sowing, irrigation, crop growth monitoring, etc.

[0017] Preferably, step S15 includes the following steps:

[0018] Step S151: Denoise the initial point cloud data of farmland to generate purified point cloud data;

[0019] Step S152: Reconstruct the point cloud surface based on the purified point cloud data to generate three-dimensional surface data of farmland;

[0020] Step S153: Extract the texture features of the farmland environment image and perform surface texture mapping on the farmland three-dimensional surface data to generate farmland three-dimensional surface mapping data.

[0021] Step S154: Integrate the three-dimensional surface mapping data of farmland into a virtual scene to generate a digital twin farmland.

[0022] This invention denoises and purifies the initial point cloud data of farmland, effectively removing noise and outliers, thus improving data accuracy and quality. This provides a clean foundation for subsequent surface reconstruction and mapping, ensuring the stability and accuracy of the modeling process. By reconstructing the point cloud surface based on the purified data, three-dimensional surface data of the farmland can be generated. This step ensures a detailed restoration of the farmland terrain, making the digital twin farmland model more realistic and operable, and providing a reliable three-dimensional spatial foundation for intelligent agricultural management. Extracting texture features from farmland environment images and mapping them onto the three-dimensional surface data enhances the visual effect of the digital twin farmland model. Texture mapping adds color, detail, and realism to the three-dimensional farmland model, helping users to more intuitively understand the actual situation of the farmland and providing more intuitive information for agricultural decision-making. Integrating the three-dimensional surface mapping data of the farmland into a virtual scene creates a complete and interactive digital twin farmland environment. This virtual scene provides users with a panoramic view of the farmland, enabling real-time simulation and observation of farmland changes, and optimizing farmland management and decision-making.

[0023] Preferably, step S2 includes the following steps:

[0024] Step S21: Obtain seed data to be sown;

[0025] Step S22: Extract the seed type, variety, and growth cycle from the seed data to be sown, classify the seed data to be sown, and generate seed classification data;

[0026] Step S23: Analyze the developmental characteristics of the seeds to be sown based on the seed classification data to obtain the developmental characteristic data of the seeds to be sown;

[0027] Step S24: Perform time series analysis on the developmental characteristics data of the seeds to be sown, extract developmental parameters at different growth stages, and generate stage-specific developmental characteristic data; import the stage-specific developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data.

[0028] This invention acquires data on seeds to be sown, accurately recording key information such as seed type, variety, and origin. This provides detailed foundational data for subsequent seed classification and developmental characteristic analysis, ensuring more scientific seed selection and management during sowing. By extracting information such as seed type, variety, and growth cycle from the seed data, seeds are precisely classified, generating seed classification data. This classification data provides a clear basis for subsequent seed management and sowing decisions, ensuring the selection of the most suitable seeds under different farmland conditions, thereby improving crop yield and quality. Developmental characteristic analysis based on seed classification data reveals the specific growth characteristics of the seeds to be sown. This developmental characteristic data includes germination rate, root development, and resistance characteristics, providing a scientific basis for formulating sowing strategies and management measures, ensuring healthy seed growth after sowing. Time-series analysis of the seed developmental characteristic data extracts key developmental parameters at different growth stages. This stage-specific data helps agricultural managers accurately grasp the needs of seeds at different growth stages, optimizing sowing timing, depth, and density, thereby improving crop growth efficiency and quality. Importing data on the phased development characteristics into digital twin farmland for sowing and growth simulation can generate more accurate sowing and growth simulation data. This allows farmland managers to rehearse the sowing and growth process in a virtual environment, adjust planting plans in a timely manner, and reduce potential risks.

[0029] Preferably, step S24, which involves importing the stage-specific developmental characteristic data into a digital twin farmland for sowing and growth simulation, includes:

[0030] The stage-specific developmental characteristics data were imported into a digital twin farmland to set sowing simulation parameters. The resulting sowing simulation parameters were: sowing depth of 3 cm, seed density of 100 seeds / m², seed germination rate of 85%, soil temperature of 20℃, and soil moisture of 0.22 cm³. 3 / cm 3 The initial seed growth rate is 1.2 mm / day, the temperature adaptability range is 18℃ to 30℃, the water requirement is 5 mm / day, the light requirement is 12 hours / day, the nutrient requirements are 10 kg / ha of nitrogen, 5 kg / ha of phosphorus, and 8 kg / ha of potassium, the crop growth cycle is 120 days, the root growth rate is 0.8 cm / day, and the crop stress resistance score is 4 points.

[0031] The stage-specific developmental characteristics data were imported into a digital twin farmland to set growth simulation parameters. The resulting growth simulation parameters were: initial growth rate of 1.5 mm / day, nutrient requirements of 15 kg / ha for nitrogen, 10 kg / ha for phosphorus, and 12 kg / ha for potassium, maximum leaf area index of 3.5, and photosynthetically active radiation requirement of 15 MJ / m². 2 / day, water evaporation rate is set at 4 mm / day, root expansion rate is 2 cm / day, growth cycle temperature requirement is between 15℃ and 30℃, growth stages are distributed as initial growth period of 30 days, rapid growth period of 60 days, maturity period of 30 days, light response coefficient is 0.9, pest and disease impact is 0.05, environmental humidity requirement is 70% to 85%, and crop maturity period is 90 days.

[0032] Agricultural simulation software was used to simulate and analyze the parameters for sowing simulation and growth simulation, generating sowing simulation data and growth simulation data.

[0033] This invention accurately simulates the sowing process under different environmental conditions by setting specific sowing parameters, such as sowing date, sowing depth, seed density, germination rate, and soil temperature and humidity. These detailed parameter settings provide precise guidance for sowing operations, ensuring that sowing depth and density meet seed growth requirements, thereby improving seed germination rate and initial growth rate. By setting key parameters such as seed germination rate and initial growth rate, the growth of seeds in different environments can be better simulated. Setting reasonable seed density, soil temperature and humidity conditions helps optimize the seed germination environment, thereby improving germination rate and initial growth speed, and reducing the failure rate in the early stages of planting. In growth simulation, setting detailed growth parameters, such as nutrient requirements, leaf area index, photosynthetically active radiation requirements, water evaporation, and root expansion rate, allows for more accurate simulation of crop needs at different growth stages. This data helps managers adjust water and fertilizer management and environmental control during the planting process to ensure optimal environmental conditions for crops at each growth stage. Precise division of growth stages (initial growth period, rapid growth period, maturity period) allows for a better understanding and prediction of different crop growth needs. Applying these phased requirements to simulations helps optimize farmland management, improve crop growth efficiency and health, and avoid resource waste. In growth simulations, settings for factors such as light response coefficient, pest and disease impact, and environmental humidity can simulate crop performance under different environmental conditions. By adjusting these parameters, light, water, and nutrient supply can be optimized in a virtual environment, ensuring healthy crop development throughout its entire growth cycle.

[0034] Preferably, the sowing profile analysis of the sowing simulation data in step S3 includes:

[0035] Extract seed location and soil temperature information from the sowing simulation data;

[0036] Based on the seed location information, the seeding simulation depth of the seeding simulation data was analyzed to obtain seed distribution data at different depths;

[0037] Soil temperature and soil moisture are extracted from the soil temperature information, and soil temperature and soil moisture are interpolated separately to generate continuous soil temperature profile data and continuous soil moisture profile data.

[0038] Based on seed distribution data, soil temperature profile data and soil moisture profile data are visualized to generate a sowing profile map.

[0039] This invention extracts seed location information from sowing simulation data, providing a clear understanding of the location and distribution of each seed. Analyzing the simulated sowing depth determines the accurate distribution of seeds in the soil, providing data support for optimizing and managing sowing depth, and helping to improve seed germination rate and early growth performance. By extracting soil temperature and humidity information and interpolating them separately, continuous soil temperature and humidity profile data can be obtained, providing detailed reference for accurately understanding the impact of soil conditions on crop growth. Temperature and humidity are key factors affecting seed germination and early growth; continuous soil temperature and humidity data helps analyze how these factors affect different stages of crop growth. Combining seed distribution data with soil temperature and humidity data allows for a comprehensive assessment of the impact of sowing depth on seed growth. At different depths, seed germination rate and growth status are affected by different temperature and humidity conditions; providing this comprehensive perspective helps optimize sowing depth, thereby improving crop health and stress resistance. By combining soil temperature profile data, soil humidity profile data, and seed distribution data, a sowing profile map is generated. Visualized sowing profiles help agricultural managers intuitively understand the impact of soil environmental conditions on seed germination and growth during the sowing process, thereby enabling better land management and crop growth regulation. Analysis of seed distribution at different depths can identify the optimal sowing depth, ensuring seeds receive the most suitable soil temperature and humidity conditions in the early stages of germination. Adjusting the sowing depth can improve seed germination rate and early growth rate, optimizing sowing effectiveness.

[0040] Preferably, the step S3, which involves analyzing the root-to-shoot ratio imbalance of the sowing profile based on growth simulation data, includes:

[0041] The morphological characteristics of the sown crops in the growth simulation data are extracted, and the growth rate of the growth simulation data is calculated based on the morphological characteristics to obtain dynamic growth data.

[0042] Based on dynamic growth data, the growth simulation data of sown crops are stratified to obtain upper and lower layer growth data; the spatial structure of the root system of sown crops is analyzed through the lower layer growth data to generate root characteristic data.

[0043] Biomass was calculated from the upper growth data to obtain canopy characteristic data;

[0044] The root-to-shoot ratio of crops is obtained by calculating the ratio between root system characteristic data and canopy characteristic data;

[0045] The root-to-shoot ratio imbalance of crops is assessed based on the sowing profile diagram, and the analysis results of the root-to-shoot ratio imbalance are generated.

[0046] This invention extracts morphological characteristics of sown crops from growth simulation data and calculates growth rates, enabling dynamic tracking of crop growth. This dynamic growth data provides crucial support for analyzing crop growth trends, cyclical changes, and growth stages, helping farmers adjust agricultural management strategies in a timely manner and ensuring appropriate resource supply to crops at different growth stages. Layering the growth simulation data of sown crops allows for separate analysis of the upper and lower growth layers. This hierarchical analysis makes management of different growth areas (such as between the ground and canopy, or between the root system and the ground) more targeted, allowing for adjustments to fertilization, irrigation, and other management measures based on the needs of different layers. Analyzing the spatial structure of the root system in the lower growth data generates root characteristic data, providing detailed information on the health and growth status of the crop roots. Roots are crucial for crops to acquire water and nutrients; accurate root characteristic analysis helps improve the precision of water and fertilizer management, ensuring adequate support for the crop roots. Biomass calculation of the upper growth data yields characteristic data of the crop canopy, reflecting the crop's photosynthetic capacity and overall growth status. Canopy biomass is a crucial indicator of crop health. High-quality canopy biomass contributes to photosynthesis, thereby increasing crop yield and stress resistance. The root-to-shoot ratio (RTR) can be calculated by comparing root and canopy characteristic data. The RTR is an important indicator of crop growth balance; both excessively high and low RTRs negatively impact crop growth and development. Calculating and adjusting the RTR can provide a more balanced growth environment, optimize resource allocation, and improve growth efficiency.

[0047] Preferably, the adjustment of the sowing simulation data based on the analysis results of the root-shoot ratio imbalance in step S3 includes:

[0048] The analysis results of root-to-shoot ratio imbalance are compared with the preset root-to-shoot ratio imbalance threshold. When the analysis results of root-to-shoot ratio imbalance are greater than or equal to the root-to-shoot ratio imbalance threshold, abnormal seed sowing data are generated.

[0049] Analyze the abnormal factors in abnormal seed sowing data to identify abnormal seed sowing factors, which include at least one of the following: abnormal seed density, abnormal sowing depth, abnormal nutrient supply, and abnormal moisture.

[0050] Adjusting seeding simulation data by addressing abnormal seed sowing factors:

[0051] If the abnormal seed sowing factor is confirmed to be abnormal seed density, then the seed density in the sowing simulation data should be increased / decreased by 10%-20%; if the abnormal seed sowing factor is confirmed to be abnormal sowing depth, then the sowing simulation data should be increased / decreased by 2-5cm; if the abnormal seed sowing factor is confirmed to be abnormal nutrient supply, then the nitrogen fertilizer supply should be increased by 10%-15% kg / ha, the phosphorus fertilizer supply should be decreased by 5%-10% kg / ha, and the potassium fertilizer supply should be increased by 10%-15% kg / ha; if the abnormal seed sowing factor is confirmed to be abnormal water, then the irrigation amount should be increased / decreased by 10%-20%, and the data should be integrated to obtain optimized sowing simulation data.

[0052] This invention accurately identifies abnormal factors in the sowing process by comparing the analysis results of root-to-shoot ratio imbalance with a preset root-to-shoot ratio imbalance threshold. Adjustments based on the root-to-shoot ratio imbalance allow for more precise determination of sowing parameters, such as seed density, sowing depth, nutrient and water supply. This helps improve sowing accuracy, avoids problems such as uneven seed distribution, excessive or insufficient nutrient supply, and optimizes sowing results. Root-to-shoot ratio imbalance often leads to uneven crop growth, affecting the resource allocation of the root system and canopy. By promptly identifying root-to-shoot ratio imbalances and adjusting sowing simulation data, root and canopy imbalances during planting can be effectively resolved, thereby enhancing the crop's adaptability to different environments. Adjusting sowing data based on the analysis results of root-to-shoot ratio imbalance provides more targeted improvements to sowing strategies. Specifically, adjusting sowing density when abnormal seed density is detected, adjusting sowing depth when abnormal sowing depth is detected, or adjusting fertilizer and irrigation amounts when abnormal nutrient and water supply are detected, such refined adjustments will greatly improve the crop planting success rate and ensure that the crop receives optimal conditions at each growth stage.

[0053] Preferably, the real-time sowing position correction of the sowing operation data in step S4 includes:

[0054] The seeding optimization simulation data is imported into the digital twin farmland for seeding path analysis to generate theoretical seeding locations;

[0055] Real-time positioning processing of the seeder is performed on the sowing operation data to generate absolute coordinate data;

[0056] The attitude of the seeding operation data is calculated based on the absolute coordinate data to generate the seeding position data of the seeder.

[0057] Compare the seeding position data of the seeder with the theoretical seeding position to generate spacing deviation data;

[0058] By using spacing deviation data to compensate for the movement of the sowing position in the sowing operation data, intelligent agricultural sowing management operations can be performed.

[0059] This invention, through real-time positioning processing and attitude calculation of the seeder, can accurately obtain the absolute position and sowing location of the seeder in the field. By comparing this with the theoretical sowing location, deviations can be detected and corrected in a timely manner, ensuring that the seeder sows accurately at the predetermined position in each operational step. This helps improve sowing accuracy and avoids the impact of errors on crop growth. This process utilizes digital twin farmland for sowing path analysis, combined with real-time positioning and attitude calculation. Through immediate processing and compensation of deviation data, the automation and intelligence level of sowing operations are improved. The automated compensation system reduces manual intervention, improves sowing efficiency and accuracy, and supports precision agriculture management. Precise sowing location ensures that each seed is in optimal soil conditions, thereby promoting uniform germination and growth. By reducing sowing location deviations, the crop growth environment is more consistent, helping to improve crop uniformity in the field, ensuring consistency in crop development stages, and contributing to increased crop yield and quality. Real-time seeding position correction can automatically identify deviations that occur during the seeding process and then dynamically adjust the seeding path. This correction enables the seeder to complete the seeding task in a shorter time and avoids reseeding or missed seeding due to deviations, thereby improving operational efficiency and resource utilization.

[0060] This specification provides an Internet of Things (IoT)-based smart agriculture management system for executing the aforementioned IoT-based smart agriculture management method. The IoT-based smart agriculture management system includes:

[0061] The 3D modeling module is used to acquire farmland geographic data and farmland environment images; extract scene feature points from the farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland;

[0062] The three-dimensional simulation module is used to acquire seed data to be sown; analyze the developmental characteristics of the seed data to be sown to obtain seed developmental characteristic data; and import the seed developmental characteristic data into a digital twin farmland for sowing and growth simulation to generate sowing simulation data and growth simulation data.

[0063] The sowing analysis module is used to perform sowing profile analysis on sowing simulation data to obtain sowing profile diagrams; it analyzes the root-to-shoot ratio imbalance of the sowing profile diagrams based on growth simulation data, and adjusts the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data;

[0064] The position correction module is used to import the seeding optimization simulation data into the smart seeder for seeding operation using the Internet of Things, and to collect seeding operation data simultaneously; it performs real-time seeding position correction on the seeding operation data to perform smart agricultural seeding management operations.

[0065] The beneficial effects of this invention lie in its ability to accurately reconstruct a three-dimensional model of farmland by acquiring farmland geographic data and farmland environmental images. This provides detailed spatial data support for crop growth and sowing activities. By combining scene feature points from environmental images with geographic data, the generated digital twin farmland accurately reflects the actual farmland's topography, landforms, and environmental conditions, providing a more intuitive and detailed visualization tool for agricultural management. Through digital twin technology, farmland status can be synchronized and updated in real time, providing high-precision data support for subsequent simulation, monitoring, and management, thereby improving the intelligence and accuracy of agricultural production. By analyzing the developmental characteristics of seeds to be sown, sowing strategies can be adjusted according to the growth requirements of different seeds, ensuring that seeds germinate and grow under optimal conditions. The generated sowing simulation data and growth simulation data provide agricultural managers with detailed crop growth models. These simulation data can help predict various stages of crop growth and provide a basis for later management. Combining seed developmental characteristic data with the digital twin farmland model allows seed developmental characteristics to match environmental conditions, thereby improving sowing efficiency and crop yield. By analyzing sowing profiles from sowing simulation data, key factors such as seed distribution in the soil and soil temperature and humidity can be revealed, providing strong data support for optimizing sowing depth and density. Analyzing the root-to-shoot ratio imbalance in sowing profiles allows for precise assessment of seed growth potential and space utilization efficiency, providing a scientific basis for optimizing sowing strategies. Adjusting sowing simulation data can effectively address root-to-shoot ratio imbalance, promoting healthy crop growth. Based on adjustments to the root-to-shoot ratio imbalance, optimized sowing simulation data can be generated, thereby improving sowing accuracy and crop growth, and providing more controllable management strategies for agricultural production. Real-time acquisition of sowing operation data and simultaneous correction of the seeder's position using IoT technology ensures the seeder maintains a precise sowing path during operation, avoiding uneven sowing caused by deviations. Real-time sowing position correction improves the automation and intelligence of sowing operations, reduces manual intervention, and increases sowing efficiency and accuracy. Precise sowing position and path control maximizes land resource utilization, ensuring seeds are sown at appropriate locations and depths, improving the spatial consistency and uniformity of crop growth. Therefore, this invention improves the accuracy and efficiency of agricultural seeding management by combining digital twin, Internet of Things and intelligent simulation technologies. Attached Figure Description

[0066] Figure 1 This is a flowchart illustrating the steps of a smart agriculture management method based on the Internet of Things.

[0067] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S1.

[0068] Figure 3 for Figure 1 A detailed flowchart illustrating the implementation steps of step S2.

[0069] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0070] 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.

[0071] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0072] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0073] To achieve the above objectives, please refer to Figures 1 to 3 A smart agriculture management method based on the Internet of Things, the method comprising the following steps:

[0074] Step S1: Acquire farmland geographic data and farmland environment images; extract scene feature points from farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland;

[0075] Step S2: Obtain seed data to be sown; analyze the developmental characteristics of the seed data to obtain seed developmental characteristic data; import the seed developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data;

[0076] Step S3: Perform sowing profile analysis on the sowing simulation data to obtain a sowing profile diagram; analyze the root-to-shoot ratio imbalance of the sowing profile diagram based on the growth simulation data, and adjust the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data;

[0077] Step S4: Use the Internet of Things to import the seeding optimization simulation data into the smart seeder for seeding operation and collect seeding operation data simultaneously; perform real-time seeding position correction on the seeding operation data to execute smart agricultural seeding management operations.

[0078] This invention, through the combination of 3D modeling and digital twin farmland, can accurately simulate the sowing and crop growth process in a virtual environment. Real-time data updates and analysis allow for fine-tuning based on environmental changes and farmland conditions. This enhanced precision ensures crops grow in the most suitable environment, maximizing yield. By acquiring and analyzing the developmental characteristics of seeds to be sown, customized sowing plans can be provided for different crop seeds, optimizing growth conditions for different crops. This personalized management improves germination rates and growth efficiency, reducing seed waste. Optimizing the root-to-shoot ratio is a key factor affecting plant health and yield. By analyzing imbalances in the root-to-shoot ratio, parameters such as sowing depth and soil fertilization can be adjusted promptly to avoid growth obstacles caused by imbalances, improving root development and leaf growth, and promoting overall healthy growth. Continuous collection and analysis of sowing simulation data, growth simulation data, and sowing operation data provide real-time decision support for agricultural managers. Specifically, growth simulation data can be used to predict crop growth trends and the likelihood of pests and diseases, allowing for proactive intervention to improve yield and quality. The introduction of intelligent seeders not only achieves precise sowing but also allows for dynamic adjustments based on real-time data. This automated operation significantly reduces human intervention and operational errors, improves sowing efficiency, and ensures crops grow at appropriate depths and spacing, contributing to even crop distribution and reducing waste and resource waste. Therefore, this invention improves the accuracy and efficiency of agricultural sowing management by combining digital twin, Internet of Things, and intelligent simulation technologies.

[0079] In this embodiment of the invention, reference Figure 1 The diagram shown illustrates the steps of a smart agriculture management method based on the Internet of Things (IoT) according to the present invention. In this example, the smart agriculture management method based on IoT includes the following steps:

[0080] Step S1: Acquire farmland geographic data and farmland environment images; extract scene feature points from farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland;

[0081] In this embodiment of the invention, high-precision geographic coordinate data of farmland is acquired through GNSS (Global Navigation Satellite System) and RTK (Real-Time Kinematics), with the measurement error controlled within ±2cm. This is combined with remote sensing imagery (0.3m resolution) to obtain large-scale farmland distribution information. Simultaneously, a UAV equipped with a multispectral camera (400–1000nm band) is used to collect farmland environmental images at an altitude of 120m, recording information such as crop growth status and surface moisture at a resolution of 5cm / pixel. Subsequently, the SIFT algorithm is used to extract feature points from the environmental images, averaging approximately 5000 feature points per image. The FLANN matching algorithm achieves over 90% accurate alignment of feature points, and spatial registration is performed using geographic coordinate information. In the 3D modeling stage of the farmland, the SfM (Structured Light Restoration) method is used to generate a 3D point cloud with an average point density of 1000 points / m². 2 Simultaneously, LiDAR scanning data was used to improve the elevation accuracy to ±5cm. Subsequently, the Poisson Surface Reconstruction algorithm was used for meshing, ultimately generating a high-precision farmland model containing 1 million to 5 million faces. Texture mapping technology was then employed to enhance the model's realism. Furthermore, soil moisture monitoring sensors (measuring range 0–60% water content, error ±1%) were integrated to achieve dynamic updates of the digital twin farmland. Finally, WebGL and Cesium.js technologies were used to visualize the model, allowing users to view farmland topography, crop status, and other information at a 1:1 real-world scale, thus generating a digital twin farmland.

[0082] Step S2: Obtain seed data to be sown; analyze the developmental characteristics of the seed data to obtain seed developmental characteristic data; import the seed developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data;

[0083] In this embodiment of the invention, basic data of the seeds to be sown are obtained, including thousand-grain weight (e.g., wheat 40±2g, corn 320±10g), seed moisture content (12%±1%), and internal void rate (≤1%) and germination embryo integrity rate (≥95%) detected by X-ray. Subsequently, near-infrared spectroscopy (900–2500nm) is used to analyze the protein content (wheat 12.5±0.8%, soybean 38±2%) and starch content (corn 70±3%) of the seeds. A 48-hour germination test is conducted at 25℃ and 60% relative humidity, and the germination rate of wheat is measured to be 92±3%, the root length growth rate is 3.2±0.5mm / day, and the shoot length growth rate is 4.5±0.7mm / day. Combining the temperature and humidity experimental data, a seed germination adaptability model is established, and the optimal growth temperature for wheat is calculated to be 18–25℃, with critical temperatures below 10℃ or above 35℃. Next, the sowing process was simulated in a digital twin farmland environment. The sowing depth for wheat was set at 3±0.5 cm, plant spacing at 5 cm, row spacing at 20 cm, and the coefficient of variation (CV) for sowing uniformity was ≤15%. Three-dimensional physical simulation was used to simulate seed settling and coverage in the soil, generating high-precision sowing point cloud data (2 cm resolution). During the growth simulation phase, the growth rate of wheat at each stage of its 120±5-day growth cycle was predicted using the WOFOST model and meteorological data (such as average temperature and precipitation over the past 5 years), for example, an average daily height increase of 1.5±0.3 cm during the jointing stage. Leaf area index (LAI) was used to simulate vegetation cover, with LAI = 0.3 at the seedling stage, 2.5 at the jointing stage, and 5.8 at the heading stage. Simultaneously, historical disease data was used to predict that the incidence of Fusarium head blight would increase to 20% under high temperature and humidity conditions. Ultimately, the system outputs sowing simulation data (sowing depth, seed distribution, sowing uniformity) and growth simulation data (growth cycle, biomass changes, environmental adaptability), providing precise decision support for intelligent farmland management.

[0084] Step S3: Perform sowing profile analysis on the sowing simulation data to obtain a sowing profile diagram; analyze the root-to-shoot ratio imbalance of the sowing profile diagram based on the growth simulation data, and adjust the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data;

[0085] In this embodiment of the invention, data such as seed sowing depth, surrounding soil structure, and water content distribution are collected using three-dimensional laser scanning technology (1 mm resolution) and high-resolution spectral imaging (spectral range 400–1000 nm). Combined with sowing point cloud data (2 cm resolution), a farmland sowing profile is generated, and key indicators such as seed location, initial root growth direction, and soil compaction are marked. The sowing depth distribution in the profile is extracted; for example, the average sowing depth of wheat seeds is 3.2 ± 0.4 cm, with a coefficient of variation (CV) of 10%. The soil moisture content around the seeds (target range 40%–60%) is calculated, and its impact on root germination is analyzed. Based on the soil profile data, the settling stability of seeds in different regions is assessed, and abnormal areas affected by soil compaction are identified. According to growth simulation data, root growth length and canopy height at different growth stages are extracted; specifically, the root length of wheat seedlings is 4.5 ± 0.5 cm, and the canopy height is 6.2 ± 0.7 cm, resulting in a root-to-shoot ratio (R / S) of 0.73. Calculate the root-to-shoot ratio (R / S) at the tillering, jointing, and heading stages, and establish growth dynamic curves. For example, R / S = 0.55 at the jointing stage and R / S = 0.48 at the heading stage. A low R / S ratio (R / S < 0.5) indicates insufficient root development, caused by shallow sowing, insufficient soil nutrients, or excessive water. A high R / S ratio (R / S > 1.0) indicates restricted above-ground growth, caused by excessive soil compaction, insufficient seed viability, or water deficit. Combining multi-dimensional data (root length, canopy height, and soil environment), classify the root-to-shoot ratio imbalance in different regions and output an imbalance distribution map. For areas with shallow sowing (< 2.5 cm), appropriately increase the sowing depth to 3.5 ± 0.3 cm to enhance root development. For areas with deep sowing (> 4.5 cm), reduce the depth to 3.0 ± 0.2 cm to avoid hindering seed emergence. Analysis of the sowing profile revealed areas with excessively high soil compaction (density > 1.6 g / cm³). 3 Adjust the pressure parameters of the sowing machinery to reduce the impact of compaction and increase the space for seed and root extension. In areas with soil moisture content below 40%, increase local irrigation to achieve soil moisture of 50% ± 5%. Based on the optimized data, readjust the seed spacing to avoid excessive competition; for example, optimize the plant spacing for wheat to 6 cm to improve ventilation and light penetration. Retest sowing in an optimized simulation environment to ensure that the coefficient of variation (CV) for sowing uniformity after adjustment is ≤ 8%. Record the adjusted sowing depth distribution, soil moisture content, compaction, and other parameters, and generate an optimized sowing profile. Calculate the optimized root-to-shoot ratio curve, aiming to stabilize the R / S ratio at the jointing stage at 0.6–0.7 and at the heading stage at 0.45–0.55, improving crop growth uniformity. Output the optimized sowing point cloud data, soil environment data, and growth model parameters to provide optimization reference for subsequent precision farmland management.

[0086] Step S4: Use the Internet of Things to import the seeding optimization simulation data into the smart seeder for seeding operation and collect seeding operation data simultaneously; perform real-time seeding position correction on the seeding operation data to execute smart agricultural seeding management operations.

[0087] In this embodiment of the invention, optimized sowing simulation data (including sowing depth, plant spacing, row spacing, soil moisture, and density) is transmitted to the intelligent seeder via a cloud platform, employing standard IoT protocols (such as MQTT or LoRa) to ensure efficient and low-latency data transmission. The formatted data is then imported into the intelligent seeder control system, including sowing path planning, sowing depth setting, and sowing speed and accuracy settings. The system supports a data reception rate of 50 data points per second, ensuring real-time performance and accuracy. Upon receiving the optimized data, the intelligent seeder automatically adjusts the sowing depth (e.g., 3.2 ± 0.4 cm), plant spacing (e.g., 6 cm), and row spacing (e.g., 20 cm), and adjusts the sowing speed and seed quantity according to soil conditions (e.g., moisture content, density). Sensors within the seeder (including depth sensors, GPS modules, humidity sensors, and temperature sensors) monitor the operating environment and machine status in real time and return the data to the control system. During the sowing operation, the intelligent seeder collects operational data in real time through integrated multiple sensors (GPS positioning, depth sensors, seed detectors, and humidity monitors). Sowing data includes: real-time sowing location (GPS coordinates), actual sowing depth, soil moisture, and operating speed. Data is collected 10 times per second to ensure accurate capture of every change during the sowing process. Operational data is synchronously uploaded to a cloud platform for real-time storage and analysis. Real-time sowing data is transmitted to a central control platform via an IoT gateway (such as a 4G / 5G module or LoRa communication) for data analysis and visualization. The system monitors every stage of the sowing operation in real time, including seed quantity, depth changes, and soil moisture deviations, ensuring operational accuracy. During operation, GPS and ground-based RTK positioning systems are used to correct the sowing location in real time. Assuming a target accuracy of ±2cm, if the sowing location deviation exceeds a threshold (e.g., ±3cm), the system will automatically adjust the seeder's trajectory. If an error in sowing depth is detected (e.g., too shallow or too deep), the system automatically controls the seeder's depth adjustment mechanism based on feedback signals from the depth sensor, ensuring the depth remains within the target range. The system performs position correction every 5 seconds to ensure accuracy throughout the entire sowing process. Based on real-time location and operational data, the central platform can monitor the progress of sowing operations in real time, ensuring that each area meets the predetermined sowing standards. Through data analysis, the platform can identify problem areas during operations (such as inconsistent sowing depth, abnormal seed application rates, etc.) and alert operators through the user interface or automatically adjust the seeder settings. During operations, the system can automatically adjust the sowing depth and application rate based on environmental factors such as soil moisture and temperature, specifically reducing the sowing depth in areas with high humidity to avoid waterlogging of the seeds. After the operation is completed, the intelligent seeder feeds back the sowing data (including operation time, sowing depth, sowing uniformity, and coverage of the operation area) to the cloud platform for data analysis and storage.By analyzing the uniformity of sowing depth (e.g., coefficient of variation (CV) ≤ 5%), the system assesses sowing quality and optimizes subsequent operational parameters. Based on data feedback, the system can generate a sowing operation report and provide optimization suggestions. Specifically, if the sowing depth is too shallow in certain areas, the system suggests increasing the seed coverage depth or adjusting the sowing speed.

[0088] As an example of the present invention, reference is made to Figure 2 As shown, in this example, step S1 includes:

[0089] Step S11: Obtain farmland geographic data using GIS technology; acquire farmland environmental images using cameras;

[0090] Step S12: Standardize the coordinates of the farmland geographic data to generate standardized farmland geographic data;

[0091] Step S13: Perform feature point detection and matching on the farmland environment image, and extract scene feature points from the farmland environment image to obtain scene feature points;

[0092] Step S14: Combine standardized farmland geographic data and scene feature points to reconstruct farmland point clouds and generate initial farmland point cloud data;

[0093] Step S15: Perform 3D surface modeling on the initial point cloud data of the farmland to generate a digital twin farmland.

[0094] In this embodiment of the invention, precise geographic data of farmland is acquired using GIS (Geographic Information System) technology via drones or satellite remote sensing. Commonly used geographic data includes farmland boundary coordinates, topographic relief, slope, land cover type, etc., with an accuracy typically within 1 meter. High-resolution remote sensing imagery (such as satellite imagery with a resolution of 30cm) is combined with ground measurement data (such as GNSS systems) to supplement the farmland geographic data, ensuring its accuracy. High-resolution cameras (such as digital cameras with a resolution of 20MP or higher) are used to photograph the farmland environment. The cameras need to be equipped with good optical zoom capabilities to adapt to different weather and lighting conditions. Multiple cameras are set up to capture images of the farmland environment from different angles to ensure full coverage. The image shooting angles must cover at least 30 degrees in each direction for subsequent feature point detection and matching. A unified coordinate reference system (such as WGS84 or UTM coordinate system) is used to transform and standardize the acquired farmland geographic data. Specifically, a projection transformation algorithm is used to convert coordinate data from various sources into a standard coordinate system to ensure data consistency. If the farmland coverage area is large, consider using a regional coordinate system (such as the geodetic coordinate system in geographic information standards) to improve the accuracy and processing efficiency of geographic data. The data accuracy target is 1-2 meters. Standardized farmland geographic data includes information such as land boundaries, topographic elevation, crop distribution, and soil type. This data will serve as the basis for subsequent 3D reconstruction and environmental analysis. Feature point detection is performed on farmland environment images using computer vision algorithms such as SIFT (Scale Invariant Feature Transform), SURF (Accelerated Robust Feature Transform), or ORB (Oriented Fast and Rotated BRIEF). These algorithms can extract feature points with unique descriptors from images, adapting to different lighting, viewing angles, and scale variations. Feature point matching algorithms (such as descriptor-based matching methods) are used to match image feature points from different viewing angles, generating a set of scene feature points containing different perspectives of the farmland environment. Based on the matching results, key scene feature points (such as farmland roads, field ridges, and crop distribution) are extracted from the farmland images. The accuracy of these feature point extractions directly affects the effect of subsequent 3D point cloud reconstruction. Robust algorithms such as RANSAC (Random Sample Consensus) are used to further filter out mismatched feature points, ensuring high accuracy of the final extracted scene feature points. The standardized farmland geographic data is combined with the extracted scene feature points, and the farmland environment is reconstructed in 3D using multi-view stereo vision (MVS) algorithms or structured light technology. During the reconstruction process, scene feature points in the farmland image are combined with elevation information from the geographic data to generate initial point cloud data for the farmland. This point cloud data includes the spatial coordinates (x, y, z) of each point, and color and density attributes of the point cloud are generated based on the image's lighting and texture information.The target point cloud density is 30-50 points per square meter to ensure sufficient detail and accuracy in the reconstructed point cloud data. Noise filtering and data sparsification are applied to the generated initial point cloud data to remove unnecessary noise points. Filters (such as median filtering and statistical filtering) are used to remove outliers caused by environmental factors (such as clouds and light reflection). 3D reconstruction software (such as MeshLab and CloudCompare) is used to perform surface modeling on the initial farmland point cloud data, generating a 3D surface model of the farmland. Surface reconstruction algorithms (such as Poisson surface reconstruction and Delaunay triangulation) are used to construct a detailed model of the farmland terrain. During this process, the model's boundaries and details are optimized by incorporating the geographical characteristics of the farmland (such as field ridges, roads, and elevation changes) to make it as close as possible to the actual farmland environment. The accuracy of the 3D surface model is required to reach 5-10 cm to ensure its usability and accuracy in the digital twin system. Based on the 3D surface modeling, a digital twin farmland system is generated. Digital twin farmland not only includes the three-dimensional geometric information of the farmland, but also data such as crop type, soil information, and water source distribution, forming a highly realistic virtual farmland model.

[0095] Preferably, step S15 includes the following steps:

[0096] Step S151: Denoise the initial point cloud data of farmland to generate purified point cloud data;

[0097] Step S152: Reconstruct the point cloud surface based on the purified point cloud data to generate three-dimensional surface data of farmland;

[0098] Step S153: Extract the texture features of the farmland environment image and perform surface texture mapping on the farmland three-dimensional surface data to generate farmland three-dimensional surface mapping data.

[0099] Step S154: Integrate the three-dimensional surface mapping data of farmland into a virtual scene to generate a digital twin farmland.

[0100] In this embodiment of the invention, point cloud denoising algorithms (such as statistical outlier removal, Voxel Grid filtering, and Bilateral filtering) are used to denoise the initial point cloud data of farmland. Point cloud data typically contains noise points, such as invalid data generated due to sensor errors, environmental interference, etc. Specific steps include: detecting and removing outliers by analyzing the local density of the neighborhood of each point, ensuring that the denoised point cloud data is more accurate. The denoising accuracy target is to remove more than 90% of the noise points, retaining approximately 80%-90% of the valid point cloud. After denoising, the resulting purified point cloud data has a more uniform spatial distribution, with the point cloud density generally controlled at around 40 points per square meter to ensure its accuracy in subsequent processing. After denoising, the data undergoes a quality check to ensure that the purified point cloud data meets the accuracy requirements, the spatial distribution of the point cloud is more uniform, and the accuracy is maintained within 5cm. Surface reconstruction algorithms (such as Poisson reconstruction, Delaunay triangular mesh reconstruction, and Alpha shape reconstruction) are used to perform surface reconstruction on the purified point cloud data. These algorithms generate a 3D surface model of farmland based on the spatial coordinates of point cloud data. The model includes features such as farmland topography, elevation changes, and crop distribution. During reconstruction, the accuracy and integrity of the 3D surface model are ensured, with a target accuracy within 10 cm. To guarantee the representation of model details, the mesh density after surface reconstruction is set at 10,000 triangular faces per square meter. Based on the surface reconstruction, the generated surface data is smoothed to eliminate rough surfaces caused by sparse or uneven point cloud data. Mesh smoothing algorithms (such as Laplacian smoothing or Taubin smoothing) are used to adjust the smoothness of the model while preserving the realistic features of the farmland topography. The optimized surface model should have good visual effects and be able to run stably in subsequent virtual scene integration. Texture features are extracted from the farmland environment images using texture analysis algorithms (such as Gabor filtering and LBP (Local Binary Pattern)). These texture analysis algorithms identify different texture regions in the farmland images, such as field ridges, crop areas, and roads. The extracted texture features are mapped onto the reconstructed 3D surface data. Texture mapping algorithms (such as UV mapping and environment mapping) are used to attach the texture information of the farmland image to the 3D surface. Through texture mapping technology, the color, lighting, and details of the farmland image are accurately transferred to the 3D surface model, enhancing the realism and visual effect of the 3D farmland model. Ultimately, the generated 3D farmland surface mapping data should contain the geometric data and surface texture data of the 3D model. This data is then integrated with other relevant data (such as soil data, moisture data, and crop distribution data) to generate a complete virtual farmland scene. During the integration process, virtual reality technology and 3D visualization technology are used to synchronize the 3D model with real-time data, enabling it to dynamically reflect changes in the farmland environment.Ultimately, by integrating 3D surface models, environmental texture data, and real-time sensor data, a complete digital twin farmland system is generated.

[0101] As an example of the present invention, reference is made to Figure 3 As shown, in this example, step S2 includes:

[0102] Step S21: Obtain seed data to be sown;

[0103] Step S22: Extract the seed type, variety, and growth cycle from the seed data to be sown, classify the seed data to be sown, and generate seed classification data;

[0104] Step S23: Analyze the developmental characteristics of the seeds to be sown based on the seed classification data to obtain the developmental characteristic data of the seeds to be sown;

[0105] Step S24: Perform time series analysis on the developmental characteristics data of the seeds to be sown, extract developmental parameters at different growth stages, and generate stage-specific developmental characteristic data; import the stage-specific developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data.

[0106] In this embodiment of the invention, machine learning models (such as K-means clustering, decision trees, support vector machines, etc.) are used to classify the seed data to be sown, extracting the seed type, variety, and growth cycle of each seed. Based on different seed growth cycles, seeds can be classified into early-maturing, mid-maturing, and late-maturing varieties. The growth cycle of each seed type varies between 60 and 150 days, categorized according to the growth habits of different crops. Based on the extracted data, a seed classification dataset is generated, containing the characteristics and growth cycle information of each seed type. The data table records the classification label, seed type, and corresponding growth cycle for each variety. Specifically, some wheat varieties have a growth cycle of 100 days, while some corn varieties have a growth cycle of 120 days. The generated seed classification data will provide a foundation for subsequent developmental characteristic analysis. Standard agricultural models (such as plant growth models, photosynthesis models, etc.) are used to analyze the developmental characteristics of different seeds. By analyzing the influence of environmental factors such as growth cycle, temperature, light, and water, the developmental speed, germination rate, and emergence period of each seed are determined. Developmental characteristics of different types and varieties of seeds were analyzed to extract relevant data, such as germination temperature range, optimal sowing depth, and suitable soil moisture. For each variety, developmental characteristic data were generated based on its growth habits. The data tables include: seed developmental characteristics for each variety (such as germination rate, emergence period, optimal growth environment, and disease resistance) and key developmental milestones (such as heading, flowering, and maturity). Specifically, a certain wheat variety achieves a 90% germination rate at 25℃, while corn varieties require 20-30℃ to ensure good germination rate and growth rate. Time series analysis methods (such as ARIMA models and periodic analysis) were used to perform time series modeling based on the developmental characteristic data to predict developmental parameters at each growth stage. By analyzing seed growth at different time points (such as 15 days, 30 days, and 60 days after sowing), developmental parameters for different growth stages were extracted, such as root length, stem height, leaf area index (LAI), and chlorophyll content. Based on time-series analysis, developmental characteristic data for different growth stages are generated. Specifically, some crops show rapid root development at 30 days and rapid leaf growth at 60 days. This generated data on stage-specific developmental characteristics is imported into a digital twin farmland system to ensure the accuracy and precision of the simulation. The digital twin farmland system can simulate the crop growth process based on its developmental characteristics. This data includes developmental parameters for each stage; specifically, some crops reach a certain root and stem length 30 days after sowing, and enter the flowering stage at 60 days. Sowing simulations are run in the digital twin farmland to simulate the impact of different sowing methods and times on crop growth. The effects of optimal sowing density, sowing depth, soil temperature, and humidity on crop growth are analyzed using the sowing simulation data. Based on the growth simulation data, crop growth outcomes under different environmental conditions can be predicted.By dynamically adjusting sowing parameters (such as sowing density and row spacing), the farmland operation process can be optimized to increase crop yield.

[0107] Preferably, step S24, which involves importing the stage-specific developmental characteristic data into a digital twin farmland for sowing and growth simulation, includes:

[0108] The stage-specific developmental characteristics data were imported into a digital twin farmland to set sowing simulation parameters. The resulting sowing simulation parameters were: sowing depth of 3 cm, seed density of 100 seeds / m², seed germination rate of 85%, soil temperature of 20℃, and soil moisture of 0.22 cm³. 3 / cm 3 The initial seed growth rate is 1.2 mm / day, the temperature adaptability range is 18℃ to 30℃, the water requirement is 5 mm / day, the light requirement is 12 hours / day, the nutrient requirements are 10 kg / ha of nitrogen, 5 kg / ha of phosphorus, and 8 kg / ha of potassium, the crop growth cycle is 120 days, the root growth rate is 0.8 cm / day, and the crop stress resistance score is 4 points.

[0109] The stage-specific developmental characteristics data were imported into a digital twin farmland to set growth simulation parameters. The resulting growth simulation parameters were: initial growth rate of 1.5 mm / day, nutrient requirements of 15 kg / ha for nitrogen, 10 kg / ha for phosphorus, and 12 kg / ha for potassium, maximum leaf area index of 3.5, and photosynthetically active radiation requirement of 15 MJ / m². 2 / day, water evaporation rate is set at 4 mm / day, root expansion rate is 2 cm / day, growth cycle temperature requirement is between 15℃ and 30℃, growth stages are distributed as initial growth period of 30 days, rapid growth period of 60 days, maturity period of 30 days, light response coefficient is 0.9, pest and disease impact is 0.05, environmental humidity requirement is 70% to 85%, and crop maturity period is 90 days.

[0110] Agricultural simulation software was used to simulate and analyze the parameters for sowing simulation and growth simulation, generating sowing simulation data and growth simulation data.

[0111] Preferably, the sowing profile analysis of the sowing simulation data in step S3 includes:

[0112] Extract seed location and soil temperature information from the sowing simulation data;

[0113] Based on the seed location information, the seeding simulation depth of the seeding simulation data was analyzed to obtain seed distribution data at different depths;

[0114] Soil temperature and soil moisture are extracted from the soil temperature information, and soil temperature and soil moisture are interpolated separately to generate continuous soil temperature profile data and continuous soil moisture profile data.

[0115] Based on seed distribution data, soil temperature profile data and soil moisture profile data are visualized to generate a sowing profile map.

[0116] In this embodiment of the invention, seed location information, including the latitude and longitude coordinates and depth information of each seed, is extracted from sowing simulation data. Seed location information is typically recorded by the GPS device of the seeder, and the data format includes the longitude and latitude of the seed location, as well as the sowing depth of each seed. Specifically, the seed location is stored as "Seed 1: Longitude 30.1234, Latitude 45.6789, Depth 2 cm," thus obtaining specific information for each seed. Soil temperature information is extracted from the sowing simulation data; this data can be acquired in real time through a sensor network. Soil temperature information includes soil temperature at different depths, typically recorded at different time points and depths. Specifically, the soil temperature is 22℃ in the 0-10 cm depth area, and 18℃ at a depth of 10-20 cm. This data allows for the determination of soil temperature distribution. Based on the seed location information, the sowing depth of each seed is calculated. By statistically analyzing the depth distribution of all seeds, the seed density at each depth level is determined. Specifically, the sowing depth can be stratified to calculate the number of seeds in depth segments such as 0-5 cm, 5-10 cm, and 10-15 cm, obtaining seed distribution data for each depth level. Based on the sowing depth analysis, seed distribution data is generated. The data format includes the number of seeds and their density in each depth segment. Specifically, the seed density in the 0-5 cm depth segment is 100 seeds / m², and the seed density in the 5-10 cm depth segment is 90 seeds / m². The extracted soil temperature data is interpolated using common interpolation methods (such as Kriging interpolation and bilinear interpolation) to generate continuous soil temperature profile data, which will help generate a smoother and more continuous soil temperature curve. Specifically, if the soil temperature data intervals are large in certain locations, continuous temperature data for those locations can be generated through interpolation. The interpolated data can display the temperature changes across the entire soil depth range; specifically: the temperature at a depth of 0-10 cm is 22℃, at 10-20 cm it is 20℃, and at 20-30 cm it is 18℃. Similarly, the extracted soil moisture data is interpolated to generate continuous soil moisture profile data. Interpolation methods such as Kriging interpolation or spline interpolation can be selected to ensure the continuity of the moisture data. Specifically, the soil moisture data shows that the moisture content is 20% at a depth of 0-10 cm and 18% at a depth of 10-20 cm. Continuous moisture data is generated through interpolation, such as 20% at 0-10 cm, 19.5% at 10-20 cm, and 18.5% at 20-30 cm. Seed distribution data is integrated with soil temperature and soil moisture profile data, and visualized according to different sowing depths. A 3D modeling tool is used to visually display seed distribution, soil temperature, and moisture at different depths. Heat maps or 3D plots are used to show the changes in soil temperature and moisture, and the seed distribution layers are marked on the plots.Specifically, it can display seed distribution, soil temperature, and humidity changes at different depths, such as 0-10 cm and 10-20 cm. Ultimately, it generates a sowing profile, showing the seed distribution at different depths, along with the corresponding soil temperature and humidity information.

[0117] Preferably, the step S3, which involves analyzing the root-to-shoot ratio imbalance of the sowing profile based on growth simulation data, includes:

[0118] The morphological characteristics of the sown crops in the growth simulation data are extracted, and the growth rate of the growth simulation data is calculated based on the morphological characteristics to obtain dynamic growth data.

[0119] Based on dynamic growth data, the growth simulation data of sown crops are stratified to obtain upper and lower layer growth data; the spatial structure of the root system of sown crops is analyzed through the lower layer growth data to generate root characteristic data.

[0120] Biomass was calculated from the upper growth data to obtain canopy characteristic data;

[0121] The root-to-shoot ratio of crops is obtained by calculating the ratio between root system characteristic data and canopy characteristic data;

[0122] The root-to-shoot ratio imbalance of crops is assessed based on the sowing profile diagram, and the analysis results of the root-to-shoot ratio imbalance are generated.

[0123] In this embodiment of the invention, morphological features of the sown crop are extracted from growth simulation data, including data on the size, volume, and distribution of the crop's stems, leaves, and roots. These morphological features are typically derived from crop growth models or through dynamic monitoring using sensors (such as image recognition). Specifically, data such as crop height, leaf area index (LAI), and root length can be extracted using imaging technology. The crop growth rate is calculated based on the extracted morphological features, typically using time-series data to calculate the rate of growth change over a certain period. The growth rate can be estimated based on changes in crop height or dry matter accumulation. Specifically, weekly growth rates are calculated by measuring crop height changes weekly, resulting in dynamic growth data. For example, if the crop grows 5 cm in height in the first week and 7 cm in the second week, the growth rate is 2 cm / week. Based on the dynamic growth data, the crop's growth is divided into upper and lower layers according to different growth stages. The upper layer typically refers to the crop's canopy (such as leaves and branches), while the lower layer refers to the root system and the surrounding soil area. Different stratification criteria are set according to the characteristics of different crops. Specifically, for maize crops, the canopy is defined as the portion from the ground to 1.5 meters, and the root system is defined as the portion from the soil surface to a depth of 1 meter. Growth data for both the upper and lower layers are extracted from growth simulation data. Upper layer growth data includes canopy height, leaf area, and number of branches, while lower layer growth data includes root depth, root length, and root density. Specifically, it is assumed that the upper leaf area of ​​the crop is 2 square meters, the root depth is 80 centimeters, and the root density is 0.4 g / cm³. Spatial structure analysis is performed on the root information in the lower layer growth data. The analysis includes indicators such as root distribution range, root depth, root quantity, and the three-dimensional spatial distribution of the roots. 3D modeling or computer vision techniques are used to analyze the spatial structure of the root system and extract its geometric features. Root structure analysis helps to understand the growth potential of the root system and its interaction with the soil environment. Through analysis, root feature data is generated, including information on root density, depth, and distribution range. Specifically, the root system is 80 cm deep in the soil, with a total root length of 10 m and a root density of 0.6 g / cm³. Based on upper-layer growth data (such as leaf area and branch length), a biomass calculation model is used to calculate the total canopy biomass. Biomass is typically estimated by measuring indicators such as leaf area and dry matter accumulation. Specifically, canopy biomass data is obtained by calculating data such as leaf area index (LAI) and photosynthetic efficiency, for example, a canopy biomass of 3 kg / m². Based on the biomass calculation results, canopy characteristic data is generated. Canopy characteristic data can include crop leaf area, canopy height, and photosynthetic efficiency. Specifically, the average leaf area of ​​the canopy is 2 m², and the canopy height is 1.5 m. Based on the root and canopy characteristic data, the root-to-shoot ratio is calculated. The root-to-shoot ratio is typically the ratio of root biomass to canopy biomass, reflecting the growth ratio of the crop's roots and canopy.Specifically, if the root biomass is 1.5 kg / m² and the canopy biomass is 3 kg / m², then the root-to-shoot ratio is 0.5. An imbalance assessment of the root-to-shoot ratio is conducted to analyze whether the crop's root-to-shoot ratio conforms to the ideal growth ratio. Imbalance usually refers to excessively weak roots or an excessively large canopy, leading to problems in water and nutrient absorption and growth. Based on the analysis results of the sowing profile, it is determined whether the crop's root-to-shoot ratio is unreasonable, such as excessively deep or shallow sowing depth restricting root growth, or uneven soil temperature and humidity affecting root growth. Based on the analysis, an analysis report on the root-to-shoot ratio imbalance is generated, including the degree of imbalance, cause analysis, and optimization suggestions.

[0124] Preferably, the adjustment of the sowing simulation data based on the analysis results of the root-shoot ratio imbalance in step S3 includes:

[0125] The analysis results of root-to-shoot ratio imbalance are compared with the preset root-to-shoot ratio imbalance threshold. When the analysis results of root-to-shoot ratio imbalance are greater than or equal to the root-to-shoot ratio imbalance threshold, abnormal seed sowing data are generated.

[0126] Analyze the abnormal factors in abnormal seed sowing data to identify abnormal seed sowing factors, which include at least one of the following: abnormal seed density, abnormal sowing depth, abnormal nutrient supply, and abnormal moisture.

[0127] Adjusting seeding simulation data by addressing abnormal seed sowing factors:

[0128] If the abnormal seed sowing factor is confirmed to be abnormal seed density, then the seed density in the sowing simulation data should be increased / decreased by 10%-20%; if the abnormal seed sowing factor is confirmed to be abnormal sowing depth, then the sowing simulation data should be increased / decreased by 2-5cm; if the abnormal seed sowing factor is confirmed to be abnormal nutrient supply, then the nitrogen fertilizer supply should be increased by 10%-15% kg / ha, the phosphorus fertilizer supply should be decreased by 5%-10% kg / ha, and the potassium fertilizer supply should be increased by 10%-15% kg / ha; if the abnormal seed sowing factor is confirmed to be abnormal water, then the irrigation amount should be increased / decreased by 10%-20%, and the data should be integrated to obtain optimized sowing simulation data.

[0129] In this embodiment of the invention, a threshold for root-to-shoot ratio imbalance is preset, typically based on historical data and crop growth patterns. This threshold can be derived experimentally or adjusted according to the growth characteristics of different crops. Specifically, the threshold can be set to indicate an abnormality when the root-to-shoot ratio imbalance is greater than 0.8. The root-to-shoot ratio imbalance analysis result obtained in step S35 is compared with the preset threshold. If the analysis result is greater than or equal to the threshold, it indicates a root-to-shoot ratio imbalance problem has occurred during sowing and must be adjusted. Specifically, if the preset threshold is 0.7 and the analysis result is 0.75, abnormal seed sowing data is generated. Once the root-to-shoot ratio imbalance analysis result exceeds the threshold, it is considered that an abnormality has occurred in the sowing process, and the system will generate abnormal seed sowing data for subsequent adjustment and optimization. The generated abnormal seed sowing data is analyzed to identify the abnormal factors causing the root-to-shoot ratio imbalance. The main abnormal factors include: abnormal seed density: seed sowing density is too high or too low. Abnormal sowing depth: seed sowing depth is too deep or too shallow. Nutrient supply anomalies: Insufficient or excessive soil nutrient supply (such as nitrogen, phosphorus, and potassium fertilizer). Moisture anomalies: Excessive or insufficient irrigation water. Based on abnormal seed sowing data, the system automatically identifies and determines the abnormal factors. Specifically, if excessive sowing density is detected, it is marked as "abnormal seed density"; if insufficient or excessive soil nutrient supply is detected, it is marked as "abnormal nutrient supply". Through analysis, the abnormal factors leading to the root-to-shoot ratio imbalance are ultimately identified, specifically: "abnormal seed density", "abnormal sowing depth", "abnormal nutrient supply", or "abnormal moisture". Based on the identified abnormal seed sowing factors, the sowing simulation data is adjusted. The adjustment method is as follows: If the abnormal factor is abnormal seed density, adjust the sowing density. The percentage increase or decrease in seed density can be set within the range of 10%-20%. Specific adjustments can be achieved by increasing or decreasing the number of seeds per hectare. Specifically, if the current sowing density is 100,000 plants / hectare, increasing it by 10% will result in a sowing density of 110,000 plants / hectare, and decreasing it by 10% will result in a sowing density of 90,000 plants / hectare. If the abnormal factor is an abnormal sowing depth, adjust the sowing depth. Increase or decrease the sowing depth by 2-5 cm. The adjustment method can be determined based on soil type, crop requirements, and climate conditions. If the current sowing depth is 10 cm, increasing it by 5 cm will result in a sowing depth of 15 cm, and decreasing it by 3 cm will result in a sowing depth of 7 cm. If the abnormal factor is an abnormal nutrient supply, adjust the soil nutrients (nitrogen, phosphorus, and potassium fertilizer). Increase nitrogen fertilizer supply by 10%-15% kg / ha, decrease phosphorus fertilizer supply by 5%-10% kg / ha, and increase potassium fertilizer supply by 10%-15% kg / ha.Specifically, if the current nitrogen fertilizer supply is 50 kg / ha, then increasing it by 15% will result in a nitrogen fertilizer supply of 57.5 kg / ha; if the current phosphate fertilizer supply is 20 kg / ha, then decreasing it by 5% will result in a phosphate fertilizer supply of 19 kg / ha; if the current potash fertilizer supply is 40 kg / ha, then increasing it by 10% will result in a potash fertilizer supply of 44 kg / ha. If the abnormal factor is abnormal water, adjust the irrigation amount. The percentage increase or decrease in irrigation amount can be set within the range of 10%-20%. Specifically, if the current irrigation amount is 500 mm, increasing it by 10% will result in an irrigation amount of 550 mm, and decreasing it by 10% will result in an irrigation amount of 450 mm. Based on the above adjustments, the optimized sowing simulation data is integrated. This data will serve as a new sowing simulation scheme for subsequent sowing operations and growth monitoring. The adjusted seed density, sowing depth, nutrient supply, and irrigation amount data are integrated into optimized sowing simulation data to provide precise guidance for actual sowing.

[0130] Preferably, the real-time sowing position correction of the sowing operation data in step S4 includes:

[0131] The seeding optimization simulation data is imported into the digital twin farmland for seeding path analysis to generate theoretical seeding locations;

[0132] Real-time positioning processing of the seeder is performed on the sowing operation data to generate absolute coordinate data;

[0133] The attitude of the seeding operation data is calculated based on the absolute coordinate data to generate the seeding position data of the seeder.

[0134] Compare the seeding position data of the seeder with the theoretical seeding position to generate spacing deviation data;

[0135] By using spacing deviation data to compensate for the movement of the sowing position in the sowing operation data, intelligent agricultural sowing management operations can be performed.

[0136] In this embodiment of the invention, adjusted and optimized sowing simulation data (such as sowing density, sowing depth, nutrient and moisture adjustments, etc.) are imported into a digital twin farmland system. The digital twin farmland is a virtual environment capable of simulating real farmland operations. Based on environmental data (such as terrain, soil type, crop type, etc.) from the digital twin farmland, the system automatically performs sowing path analysis. This analysis takes into account the actual conditions of the farmland, generating an ideal sowing path and the theoretical sowing position for each seed. Through sowing path analysis, the system generates the theoretical sowing position for each seed, which serves as a reference point in the sowing operation for subsequent correction of the seeder's position. The seeder is located in real time using positioning technologies such as GPS systems and inertial measurement units (IMUs). The real-time position of the seeder in the farmland is acquired through GPS and sensor systems. After calculation, the real-time positioning information of the seeder generates absolute coordinate data (usually based on longitude, latitude, and altitude). This data represents the actual position of the seeder in the farmland. Attitude calculation refers to determining the specific orientation and posture of the seeder during operation by analyzing its attitude (such as pitch angle, roll angle, and yaw angle). This step typically utilizes IMU sensors and related algorithms (such as Kalman filtering) for calculation. Based on the seeder's absolute coordinates and attitude calculation results, precise sowing position data is calculated, including the seeder's specific location at each time point and its trajectory in the field. The actual sowing position (the seeder position data generated in step S43) is compared with the theoretical sowing position (the theoretical sowing position generated in step S41). The theoretical sowing position is an ideal position derived from sowing optimization simulation data and path analysis in a digital twin farmland, while the actual sowing position is acquired in real time by the positioning system. Based on the difference between the actual and theoretical sowing positions, the deviation of the sowing position (i.e., the distance or angular difference between the actual and theoretical positions) is calculated, which can be done using methods such as Euclidean distance, Manhattan distance, or angular difference. The sowing position deviation is converted into spacing deviation data, which can reveal displacement errors or inaccuracies that occur during the sowing process. Specifically, if the deviation between the seeder and the theoretical position exceeds a preset range, it indicates a problem in the sowing operation, affecting crop uniformity. Based on the spacing deviation data, the sowing operation data is dynamically adjusted. The seeder's travel path and sowing position are adjusted by compensating for the seeder's motion error. Compensation methods include: correcting the seeder's forward direction or speed based on the deviation data to ensure the seeder travels along the predetermined sowing path; and adjusting the seeder's control system to correct the sowing position in real time, ensuring that the sowing position of each seed meets the theoretical requirements. The compensated sowing operation data is then used to execute subsequent sowing management tasks.Through intelligent agricultural systems, precision sowing is implemented to improve the uniformity of crop growth and optimize resource allocation (such as water, nutrients, and seed density). The effects of adjustments to sowing operations are fed back in real time, and sowing paths and strategies are continuously optimized based on the actual results of each sowing, forming a closed-loop feedback mechanism for intelligent agriculture.

[0137] Therefore, the embodiments should be considered 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.

[0138] 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 present 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 smart agriculture management method based on the Internet of Things, characterized in that, Includes the following steps: Step S1: Acquire farmland geographic data and farmland environment images; extract scene feature points from farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland; Step S2: Obtain seed data to be sown; analyze the developmental characteristics of the seed data to obtain seed developmental characteristic data; import the seed developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data; Step S3: Perform sowing profile analysis on the sowing simulation data to obtain a sowing profile diagram; analyze the root-to-shoot ratio imbalance of the sowing profile diagram based on the growth simulation data, and adjust the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data; wherein, the analysis of the root-to-shoot ratio imbalance of the sowing profile diagram based on the growth simulation data in step S3 includes: The morphological characteristics of the sown crops in the growth simulation data are extracted, and the growth rate of the growth simulation data is calculated based on the morphological characteristics to obtain dynamic growth data. Based on dynamic growth data, the growth simulation data of sown crops are stratified to obtain upper and lower layer growth data; the spatial structure of the root system of sown crops is analyzed through the lower layer growth data to generate root characteristic data. Biomass was calculated from the upper growth data to obtain canopy characteristic data; The root-to-shoot ratio of crops is obtained by calculating the ratio between root system characteristic data and canopy characteristic data; The crop root-to-shoot ratio imbalance is assessed based on the sowing profile, and analysis results of the root-to-shoot ratio imbalance are generated; wherein, the adjustment of the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance in step S3 includes: The analysis results of root-to-shoot ratio imbalance are compared with the preset root-to-shoot ratio imbalance threshold. When the analysis results of root-to-shoot ratio imbalance are greater than or equal to the root-to-shoot ratio imbalance threshold, abnormal seed sowing data are generated. Analyze the abnormal factors in abnormal seed sowing data to identify abnormal seed sowing factors, which include at least one of the following: abnormal seed density, abnormal sowing depth, abnormal nutrient supply, and abnormal moisture. By adjusting the seed sowing simulation data through abnormal seed sowing factors, optimized sowing simulation data is obtained, wherein the adjustment includes increasing or decreasing; wherein, the sowing profile analysis of the sowing simulation data in step S3 includes: Extract seed location and soil temperature information from the sowing simulation data; Based on the seed location information, the seeding simulation depth of the seeding simulation data was analyzed to obtain seed distribution data at different depths; Soil temperature and soil moisture are extracted from the soil temperature information, and soil temperature and soil moisture are interpolated separately to generate continuous soil temperature profile data and continuous soil moisture profile data. Based on seed distribution data, soil temperature profile data and soil moisture profile data are visualized to generate a sowing profile map; Step S4: Use the Internet of Things to import the seeding optimization simulation data into the smart seeder for seeding operation and collect seeding operation data simultaneously; perform real-time seeding position correction on the seeding operation data to execute smart agricultural seeding management operations.

2. The smart agriculture management method based on the Internet of Things according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Obtain farmland geographic data using GIS technology; acquire farmland environmental images using cameras; Step S12: Standardize the coordinates of the farmland geographic data to generate standardized farmland geographic data; Step S13: Perform feature point detection and matching on the farmland environment image, and extract scene feature points from the farmland environment image to obtain scene feature points; Step S14: Combine standardized farmland geographic data and scene feature points to reconstruct farmland point clouds and generate initial farmland point cloud data; Step S15: Perform 3D surface modeling on the initial point cloud data of the farmland to generate a digital twin farmland.

3. The smart agriculture management method based on the Internet of Things according to claim 2, characterized in that, Step S15 includes the following steps: Step S151: Denoise the initial point cloud data of farmland to generate purified point cloud data; Step S152: Reconstruct the point cloud surface based on the purified point cloud data to generate three-dimensional surface data of farmland; Step S153: Extract the texture features of the farmland environment image and perform surface texture mapping on the farmland three-dimensional surface data to generate farmland three-dimensional surface mapping data. Step S154: Integrate the three-dimensional surface mapping data of farmland into a virtual scene to generate a digital twin farmland.

4. The smart agriculture management method based on the Internet of Things according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Obtain seed data to be sown; Step S22: Extract the seed type, variety, and growth cycle from the seed data to be sown, classify the seed data to be sown, and generate seed classification data; Step S23: Analyze the developmental characteristics of the seeds to be sown based on the seed classification data to obtain the developmental characteristic data of the seeds to be sown; Step S24: Perform time series analysis on the developmental characteristics data of the seeds to be sown, extract developmental parameters at different growth stages, and generate stage-specific developmental characteristic data; import the stage-specific developmental characteristic data into the digital twin farmland for sowing simulation and growth simulation, and generate sowing simulation data and growth simulation data.

5. The smart agriculture management method based on the Internet of Things according to claim 4, characterized in that, Step S24, which involves importing the stage-specific developmental characteristic data into a digital twin farmland for sowing and growth simulation, includes: The stage-specific developmental characteristics data are imported into the digital twin farmland to set the sowing simulation parameters, thus obtaining the sowing simulation setting parameters; The phased developmental characteristic data are imported into the digital twin farmland to set growth simulation parameters, thus obtaining the growth simulation setting parameters; Agricultural simulation software was used to simulate and analyze the parameters for sowing simulation and growth simulation, generating sowing simulation data and growth simulation data.

6. The smart agriculture management method based on the Internet of Things according to claim 1, characterized in that, Step S4, which involves real-time correction of the sowing position in the sowing operation data, includes: The seeding optimization simulation data is imported into the digital twin farmland for seeding path analysis to generate theoretical seeding locations; Real-time positioning processing of the seeder is performed on the sowing operation data to generate absolute coordinate data; The attitude of the seeding operation data is calculated based on the absolute coordinate data to generate the seeding position data of the seeder. Compare the seeding position data of the seeder with the theoretical seeding position to generate spacing deviation data; By using spacing deviation data to compensate for the movement of the sowing position in the sowing operation data, intelligent agricultural sowing management operations can be performed.

7. A smart agricultural management system based on the Internet of Things, characterized in that, For executing the IoT-based smart agriculture management method as described in claim 1, the IoT-based smart agriculture management system includes: The 3D modeling module is used to acquire farmland geographic data and farmland environment images; extract scene feature points from the farmland environment images, and combine them with farmland geographic data to perform 3D modeling of farmland and generate digital twin farmland; The three-dimensional simulation module is used to acquire seed data to be sown; analyze the developmental characteristics of the seed data to be sown to obtain seed developmental characteristic data; and import the seed developmental characteristic data into a digital twin farmland for sowing and growth simulation to generate sowing simulation data and growth simulation data. The sowing analysis module is used to perform sowing profile analysis on sowing simulation data to obtain sowing profile diagrams; it analyzes the root-to-shoot ratio imbalance of the sowing profile diagrams based on growth simulation data, and adjusts the sowing simulation data based on the analysis results of the root-to-shoot ratio imbalance to obtain optimized sowing simulation data; The position correction module is used to import the seeding optimization simulation data into the smart seeder for seeding operation using the Internet of Things, and to collect seeding operation data simultaneously; it performs real-time seeding position correction on the seeding operation data to perform smart agricultural seeding management operations.

Images