Field crop full-cycle planting management system and method based on digital twinning

Abstract

The invention discloses a field crop full-period planting management system and method based on digital twinning. Comprising a multi-source sensing and data acquisition module, a crop-environment digital twinborn modeling module, a full-cycle growth state evaluation module, an intelligent planting decision generation module, a precise execution and process control module, a dynamic growth prediction module and a twinborn model iteration and optimization module. According to the method, a space-time alignment algorithm and a mapping precision function are introduced, dynamic association of physical and virtual environments is constructed, it is ensured that a virtual model can accurately reflect the real state of crops and the environment, and a reliable basis is provided for management decision making; a management scheme adaptive to a growth period is deduced and generated based on twinborn simulation, advanced intervention is realized in combination with dynamic prediction, decision logic is continuously optimized through a model iteration mechanism, and scientificity and adaptability of decision are improved.

Description

Technical Field

[0001] This invention relates to the field of digital agriculture technology, specifically to a field crop full-cycle planting management system and method based on digital twins. Background Technology

[0002] With the acceleration of agricultural modernization, the demand for precision and intelligent management in field crop cultivation is becoming increasingly urgent. Current crop management systems mostly adopt a single data monitoring and fixed decision-making model, which suffers from the following problems: a lack of dynamic correlation between the physical and virtual environments, making it difficult to achieve real-time mapping of crop growth status; management decisions rely on experience or static models, failing to respond to dynamic environmental changes; fragmented data throughout the entire cycle, insufficient coordination among planting stages, and difficulty in addressing the management needs of different growth stages; and a lack of closed-loop optimization mechanisms, limiting the system's adaptability and robustness. Therefore, there is an urgent need to construct a full-cycle field crop cultivation management system and methodology based on digital twins to achieve digital, visualized, and intelligent control of the planting process. Summary of the Invention

[0003] The purpose of this invention is to provide a digital twin-based full-cycle crop planting management system and method to solve the problems existing in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a field crop full-cycle planting management system based on digital twins, including a multi-source sensing and data acquisition module, a crop-environment digital twin modeling module, a full-cycle growth status assessment module, an intelligent planting decision generation module, a precision execution and process control module, a dynamic growth prediction module, and a twin model iteration and optimization module;

[0005] The multi-source sensing and data acquisition module is based on a distributed sensor network and an integrated air-space-ground monitoring device to realize the synchronous acquisition and preprocessing of field environment and crop physiological data.

[0006] The crop-environment digital twin modeling module constructs a virtual mapping model of crops and planting environment, and achieves dynamic synchronization between physical entities and virtual models through a spatiotemporal alignment algorithm;

[0007] The full-cycle growth status assessment module establishes a multi-dimensional assessment index system based on the output data of the twin model, and quantifies the crop growth health and potential.

[0008] The intelligent planting decision generation module combines growth assessment results with twin simulation to generate personalized planting management plans that are adapted to the growth period.

[0009] The precision execution and process control module translates decision commands into actions of the execution equipment and ensures operational accuracy through a closed-loop feedback mechanism.

[0010] The dynamic growth prediction module integrates historical data and real-time parameters to establish a crop growth trend prediction model, enabling the prediction of growth period and yield potential.

[0011] The twin model iteration and optimization module is based on full-cycle data accumulation and continuously optimizes the twin model parameters and decision logic through machine learning algorithms.

[0012] Preferably, the multi-source sensing and data acquisition module includes a distributed sensing unit, a space-air monitoring unit, a data preprocessing unit, and a data fusion unit;

[0013] Distributed sensing unit: Composed of soil parameter sensors, atmospheric environment sensors and crop physiological sensors, it realizes the synchronous acquisition of data on soil moisture, atmospheric temperature and humidity and crop chlorophyll content.

[0014] Space-Air Monitoring Unit: Integrates UAV remote sensing equipment and satellite remote sensing data interface to acquire macroscopic data on vegetation index and canopy temperature at the field scale;

[0015] Data preprocessing unit: Kalman filtering algorithm is used to eliminate sensor noise, Lagrange interpolation is used to fill in missing data values, and time series data is smoothed.

[0016] Data fusion unit: Based on DS evidence theory, it fuses multi-source data to establish a spatiotemporally consistent data set, providing input for twin modeling.

[0017] Preferably, the crop-environment digital twin modeling module includes a geometric modeling unit, a physical modeling unit, a spatiotemporal alignment unit, and a model accuracy verification unit;

[0018] Geometric Modeling Unit: Based on UAV point cloud data and satellite imagery, construct a three-dimensional geometric model of field topography and crop canopy, preserving key features of field boundaries and irrigation facilities;

[0019] Physical modeling unit: Establish crop physiological growth model and environmental action model, integrating the physiological processes of photosynthetic rate and respiration consumption with the coupling relationship of environmental factors such as light and water;

[0020] Spatiotemporal alignment unit: Employs a spatiotemporal calibration algorithm to achieve temporal synchronization and spatial matching between multi-source data and virtual models, establishing a one-to-one mapping relationship between physical entities and virtual models;

[0021] Model accuracy verification unit: Calculates the deviation between the virtual model and the physical entity using a mapping accuracy function, the formula of which is:

[0022]

[0023] Where M is the comprehensive mapping accuracy of the twin model; m is the number of parameter types involved in the validation (such as soil moisture, plant height, chlorophyll content, etc.); ω k Let be the weight coefficients of the k-th class parameter, satisfying The parameters are set according to their degree of influence on growth; V k P represents the k-th type of parameter value output by the virtual model. k The k-th type of parameter value is the actual measured value of the physical entity.

[0024] Preferably, the full-cycle growth status assessment module includes an indicator construction unit, a data normalization unit, a health calculation unit, and a status grading unit;

[0025] Indicator construction unit: Establish a three-level evaluation indicator system covering environmental suitability, physiological health, and growth consistency. Environmental suitability includes water availability and light availability; physiological health includes chlorophyll content and plant height growth rate; and growth consistency includes the coefficient of variation of plant height in the population.

[0026] Data normalization unit: The min-max normalization method is used to convert indicators of different dimensions into dimensionless values ​​in the range [0,1], thereby eliminating the difference in magnitude between indicators;

[0027] Health Calculation Unit: The crop growth health is calculated using a weighted summation formula, which is:

[0028]

[0029] Where H represents crop growth health, with a value ranging from [0,1], and a larger value indicates better growth status; n represents the number of primary indicators; p i Let α be the number of secondary indicators under the i-th primary indicator; i The weight of the i-th primary indicator satisfies β ij Let the weight of the j-th secondary indicator under the i-th primary indicator satisfy the following condition: X ij This is the normalized value of the j-th secondary indicator under the i-th primary indicator;

[0030] Status grading unit: Growth status is divided into levels based on health H: H≥0.8 is excellent, 0.6≤H<0.8 is good, 0.4≤H<0.6 is average, and H<0.4 is poor. Corresponding focus points are generated for different levels.

[0031] Preferably, the intelligent planting decision generation module includes a growth period identification unit, a simulation and deduction unit, a scheme generation unit, and a scheme optimization unit;

[0032] Growth stage identification unit: Based on crop plant type parameters and phenological characteristics in the twin model, combined with the accumulated temperature model, the crop growth stage (such as sowing stage, tillering stage, grain filling stage, harvest stage, etc.) is automatically identified.

[0033] Simulation and extrapolation unit: Input different management measures parameters (such as irrigation amount, fertilizer amount, pesticide application time, etc.) into the virtual model to extrapolate the changing trends of crop growth status under different schemes;

[0034] Solution generation unit: With the goal of maximizing growth health and minimizing resource consumption, it generates an initial planting management plan, specifying the implementation time, location, and parameters of each measure;

[0035] Solution optimization unit: Based on the Pareto optimality principle, the initial solution is optimized in multiple objectives, conflicting terms are eliminated, and the final executable solution is formed.

[0036] Preferably, the precision execution and process control module includes an equipment scheduling unit, an action analysis unit, an execution monitoring unit, and a closed-loop correction unit;

[0037] Equipment scheduling unit: Based on the management plan and equipment distribution information, it uses a greedy algorithm to allocate work tasks and determine the work sequence and path of irrigation machines, fertilizer machines, drones or other equipment;

[0038] Action parsing unit: converts the management parameters in the scheme into control commands that the equipment can recognize, such as the flow adjustment command of the irrigation machine and the fertilizer ratio command of the fertilizer applicator.

[0039] Execution monitoring unit: Collects operation parameters in real time through equipment sensors, feeds them back to the twin model, and monitors whether the operation process meets the requirements of the plan;

[0040] Closed-loop correction unit: When the deviation between the operation parameters and the plan exceeds the threshold, the equipment control commands are automatically adjusted to correct the operation deviation and ensure execution accuracy.

[0041] Preferably, the dynamic growth prediction module includes a sample construction unit, a prediction model training unit, a prediction execution unit, and an error correction unit;

[0042] Sample construction unit: Using historical planting data and real-time monitoring data as inputs, and future crop height and yield potential as outputs, a training sample set for the prediction model is constructed.

[0043] Prediction model training unit: The growth prediction model is trained using a Long Short-Term Memory (LSTM) network, which integrates time series features and environmental influencing factors;

[0044] Prediction Execution Unit: Inputs current data into the trained model and outputs a prediction of crop growth status over a certain future period. The prediction formula is as follows:

[0045] Y t+τ =σ(W1X t +W2H t +b)

[0046] Among them, Y t+τ σ represents the predicted crop growth value (e.g., plant height, yield) at time τ; σ is the activation function, using the ReLU function; W1 is the input layer weight matrix; X t W1 is the input feature vector at time t, including but not limited to environmental parameters and physiological parameters; W2 is the hidden layer weight matrix; H t Let be the hidden layer state at time t; b is the bias term.

[0047] Error correction unit: It adopts a rolling prediction mechanism to periodically compare the measured data with the predicted data, correct the model parameters, and reduce the prediction error.

[0048] Preferably, the twin model iteration and optimization module includes a data storage unit, a parameter optimization unit, a logic update unit, and a model verification unit;

[0049] Data storage unit: A time-series database is used to store monitoring data, execution data, prediction data and evaluation data throughout the entire cycle, and to establish a planting data archive;

[0050] Parameter optimization unit: Based on the gradient descent algorithm, with the goal of minimizing the model prediction error, it continuously optimizes the physiological parameters and environmental coupling parameters in the twin model;

[0051] Logical update unit: Based on new planting experience and data feedback, update the weights of growth assessment indicators and the logical parameters of decision generation rules;

[0052] Model Validation Unit: The performance of the optimized model is validated using new planting cycle data to ensure improved model accuracy and adaptability.

[0053] The planting management method based on a digital twin-based full-cycle planting management system for field crops includes the following steps:

[0054] Step 1: System Initialization and Twin Map Construction

[0055] The multi-source sensing and data acquisition module is activated, the sensor network and monitoring equipment parameters are configured, and the initial topographic data, basic soil parameters and crop variety information of the field are collected. The crop-environment digital twin modeling module builds a three-dimensional geometric model and a basic physical model of the field based on the initial data, completes the system initialization, and enters standby mode.

[0056] Step 2: Full-cycle data acquisition and preprocessing

[0057] The distributed sensing unit and the aerospace monitoring unit collect multi-source data on soil moisture, atmospheric environment, and crop physiology at a set cycle; the data preprocessing unit eliminates noise and fills in missing values; the data fusion unit realizes the spatiotemporal alignment of multi-source data, forms a standardized data set, and transmits it to the twin modeling module in real time.

[0058] Step 3: Twin Model Update and Growth Assessment

[0059] The crop-environment digital twin modeling module updates the virtual model parameters based on standardized data and verifies the consistency between the model and the physical entity through a mapping accuracy function; the full-cycle growth status assessment module calls the updated model data, calculates the growth health, classifies the growth status level, and outputs an assessment report.

[0060] Step 4: Intelligent Planting Decision Generation and Optimization

[0061] The intelligent planting decision generation module identifies the current growth stage of the crop and simulates the implementation effects of different management schemes in a virtual model. Based on the health assessment results, it generates and optimizes planting management schemes, clarifying the specific parameters and implementation requirements of irrigation, fertilization, and pesticide application measures.

[0062] Step 5: Precise Execution and Process Control

[0063] The precision execution and process control module schedules the corresponding work equipment and parses the optimized decision plan into equipment control commands. During the equipment's operation, the module monitors the work parameters in real time and adjusts the commands through a closed-loop correction mechanism to ensure that the operation is executed according to the plan.

[0064] Step 6: Dynamic Growth Prediction and Intervention

[0065] The dynamic growth prediction module predicts future crop growth trends based on current data and the trained model. If the prediction results indicate that the growth status may decline to an unfavorable level, the decision-making module is triggered to regenerate the intervention plan and take control measures in advance.

[0066] Step 7: Full-cycle data accumulation and model iteration

[0067] The twin model iteration and optimization module stores various types of data throughout the entire cycle, forming a planting data archive; it uses optimization algorithms to adjust the twin model parameters and decision logic, verifies the model performance with new data, completes model iteration, and provides an optimized foundation for the next season's planting management.

[0068] Compared with the prior art, the beneficial effects of the present invention are:

[0069] This invention achieves seamless data integration and closed-loop management throughout the entire process from sowing to harvest, breaking down information silos in the planting process and improving the systematicness and coherence of planting management. It introduces a spatiotemporal alignment algorithm and mapping accuracy function to construct a dynamic relationship between the physical and virtual environments, ensuring that the virtual model accurately reflects the real state of crops and the environment, providing a reliable basis for management decisions. Based on twin simulation, it generates management plans adapted to the growth period, combining dynamic prediction to achieve early intervention. Simultaneously, it continuously optimizes the decision-making logic through a model iteration mechanism, improving the scientific rigor and adaptability of the decisions. Through equipment scheduling optimization and a closed-loop correction mechanism, it ensures the precise implementation of management plans, reduces human error, and improves the standardization of planting operations. Attached Figure Description

[0070] Figure 1 This is a system module diagram of the present invention;

[0071] Figure 2 This is a schematic diagram of the multi-source sensing and data acquisition module of the present invention;

[0072] Figure 3 This is a schematic diagram of the intelligent planting decision generation module of the present invention;

[0073] Figure 4 This is a schematic diagram of the precision execution and process control module of the present invention;

[0074] Figure 5 This is a flowchart of the method of the present invention. Detailed Implementation

[0075] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0076] Please see Figure 1-5 This invention provides a field crop full-cycle planting management system based on digital twins, including a multi-source sensing and data acquisition module, a crop-environment digital twin modeling module, a full-cycle growth status assessment module, an intelligent planting decision generation module, a precision execution and process control module, a dynamic growth prediction module, and a twin model iteration and optimization module.

[0077] The multi-source sensing and data acquisition module is based on a distributed sensor network and an integrated air-space-ground monitoring device to realize the synchronous acquisition and preprocessing of field environment and crop physiological data;

[0078] The crop-environment digital twin modeling module constructs a virtual mapping model of crops and planting environment, and achieves dynamic synchronization between physical entities and virtual models through a spatiotemporal alignment algorithm;

[0079] The full-cycle growth status assessment module, based on the output data of the twin model, establishes a multi-dimensional assessment index system to quantify the crop growth health and potential.

[0080] The intelligent planting decision generation module combines growth assessment results with twin simulation to generate personalized planting management plans that are adapted to the growth period.

[0081] The precision execution and process control module translates decision-making instructions into actions of the executing equipment, and ensures operational accuracy through a closed-loop feedback mechanism.

[0082] The dynamic growth prediction module integrates historical data and real-time parameters to establish a crop growth trend prediction model, enabling the prediction of growth period and yield potential.

[0083] The twin model iteration and optimization module is based on full-cycle data accumulation and continuously optimizes the twin model parameters and decision logic through machine learning algorithms.

[0084] The multi-source sensing and data acquisition module includes a distributed sensing unit, a space-air monitoring unit, a data preprocessing unit, and a data fusion unit;

[0085] Distributed sensing unit: Composed of soil parameter sensors, atmospheric environment sensors and crop physiological sensors, it realizes the synchronous acquisition of data on soil moisture, atmospheric temperature and humidity and crop chlorophyll content.

[0086] Space-Air Monitoring Unit: Integrates UAV remote sensing equipment and satellite remote sensing data interface to acquire macroscopic data on vegetation index and canopy temperature at the field scale;

[0087] Data preprocessing unit: Kalman filtering algorithm is used to eliminate sensor noise, Lagrange interpolation is used to fill in missing data values, and time series data is smoothed.

[0088] Data fusion unit: Based on DS evidence theory, it fuses multi-source data to establish a spatiotemporally consistent data set, providing input for twin modeling.

[0089] The crop-environment digital twin modeling module includes a geometric modeling unit, a physical modeling unit, a spatiotemporal alignment unit, and a model accuracy verification unit;

[0090] Geometric Modeling Unit: Based on UAV point cloud data and satellite imagery, construct a three-dimensional geometric model of field topography and crop canopy, preserving key features of field boundaries and irrigation facilities;

[0091] Physical modeling unit: Establish crop physiological growth model and environmental action model, integrating the physiological processes of photosynthetic rate and respiration consumption with the coupling relationship of environmental factors such as light and water;

[0092] Spatiotemporal alignment unit: Employs a spatiotemporal calibration algorithm to achieve temporal synchronization and spatial matching between multi-source data and virtual models, establishing a one-to-one mapping relationship between physical entities and virtual models;

[0093] Model accuracy verification unit: Calculates the deviation between the virtual model and the physical entity using a mapping accuracy function, the formula of which is:

[0094]

[0095] Where M is the comprehensive mapping accuracy of the twin model; m is the number of parameter types involved in the validation (such as soil moisture, plant height, chlorophyll content, etc.); ω k Let be the weight coefficients of the k-th class parameter, satisfying The parameters are set according to their degree of influence on growth; V k P represents the k-th type of parameter value output by the virtual model. k The k-th type of parameter value is the actual measured value of the physical entity.

[0096] The full-cycle growth status assessment module includes an indicator construction unit, a data normalization unit, a health calculation unit, and a status grading unit.

[0097] Indicator construction unit: Establish a three-level evaluation indicator system covering environmental suitability, physiological health, and growth consistency. Environmental suitability includes water availability and light availability; physiological health includes chlorophyll content and plant height growth rate; and growth consistency includes the coefficient of variation of plant height in the population.

[0098] Data normalization unit: The min-max normalization method is used to convert indicators of different dimensions into dimensionless values ​​in the range [0,1], thereby eliminating the difference in magnitude between indicators;

[0099] Health Calculation Unit: The crop growth health is calculated using a weighted summation formula, which is:

[0100]

[0101] Where H represents crop growth health, with a value ranging from [0,1], and a larger value indicates better growth status; n represents the number of primary indicators; p i Let α be the number of secondary indicators under the i-th primary indicator; i The weight of the i-th primary indicator satisfies β ij Let the weight of the j-th secondary indicator under the i-th primary indicator satisfy the following condition: X ijThis is the normalized value of the j-th secondary indicator under the i-th primary indicator;

[0102] Status grading unit: Growth status is divided into levels based on health H: H≥0.8 is excellent, 0.6≤H<0.8 is good, 0.4≤H<0.6 is average, and H<0.4 is poor. Corresponding focus points are generated for different levels.

[0103] The intelligent planting decision generation module includes a growth period identification unit, a simulation and deduction unit, a scheme generation unit, and a scheme optimization unit;

[0104] Growth stage identification unit: Based on crop plant type parameters and phenological characteristics in the twin model, combined with the accumulated temperature model, the crop growth stage (such as sowing stage, tillering stage, grain filling stage, harvest stage, etc.) is automatically identified.

[0105] Simulation and extrapolation unit: Input different management measures parameters (such as irrigation amount, fertilizer amount, pesticide application time, etc.) into the virtual model to extrapolate the changing trends of crop growth status under different schemes;

[0106] Solution generation unit: With the goal of maximizing growth health and minimizing resource consumption, it generates an initial planting management plan, specifying the implementation time, location, and parameters of each measure;

[0107] Solution optimization unit: Based on the Pareto optimality principle, the initial solution is optimized in multiple objectives, conflicting terms are eliminated, and the final executable solution is formed.

[0108] The precision execution and process control module includes an equipment scheduling unit, an action analysis unit, an execution monitoring unit, and a closed-loop correction unit;

[0109] Equipment scheduling unit: Based on the management plan and equipment distribution information, it uses a greedy algorithm to allocate work tasks and determine the work sequence and path of irrigation machines, fertilizer machines, drones or other equipment;

[0110] Action parsing unit: converts the management parameters in the scheme into control commands that the equipment can recognize, such as the flow adjustment command of the irrigation machine and the fertilizer ratio command of the fertilizer applicator.

[0111] Execution monitoring unit: Collects operation parameters in real time through equipment sensors, feeds them back to the twin model, and monitors whether the operation process meets the requirements of the plan;

[0112] Closed-loop correction unit: When the deviation between the operation parameters and the plan exceeds the threshold, the equipment control commands are automatically adjusted to correct the operation deviation and ensure execution accuracy.

[0113] The dynamic growth prediction module includes a sample construction unit, a prediction model training unit, a prediction execution unit, and an error correction unit;

[0114] Sample construction unit: Using historical planting data and real-time monitoring data as inputs, and future crop height and yield potential as outputs, a training sample set for the prediction model is constructed.

[0115] Prediction model training unit: The growth prediction model is trained using a Long Short-Term Memory (LSTM) network, which integrates time series features and environmental influencing factors;

[0116] Prediction Execution Unit: Inputs current data into the trained model and outputs a prediction of crop growth status over a certain future period. The prediction formula is as follows:

[0117] Y t+τ =σ(W1X t +W2H t +b)

[0118] Among them, Y t+τ σ represents the predicted crop growth value (e.g., plant height, yield) at time τ; σ is the activation function, using the ReLU function; W1 is the input layer weight matrix; X t W1 is the input feature vector at time t, including but not limited to environmental parameters and physiological parameters; W2 is the hidden layer weight matrix; H t Let be the hidden layer state at time t; b is the bias term.

[0119] Error correction unit: It adopts a rolling prediction mechanism to periodically compare the measured data with the predicted data, correct the model parameters, and reduce the prediction error.

[0120] The twin model iteration and optimization module includes a data storage unit, a parameter optimization unit, a logic update unit, and a model verification unit;

[0121] Data storage unit: A time-series database is used to store monitoring data, execution data, prediction data and evaluation data throughout the entire cycle, and to establish a planting data archive;

[0122] Parameter optimization unit: Based on the gradient descent algorithm, with the goal of minimizing the model prediction error, it continuously optimizes the physiological parameters and environmental coupling parameters in the twin model;

[0123] Logical update unit: Based on new planting experience and data feedback, update the weights of growth assessment indicators and the logical parameters of decision generation rules;

[0124] Model Validation Unit: The performance of the optimized model is validated using new planting cycle data to ensure improved model accuracy and adaptability.

[0125] The planting management method based on a digital twin-based full-cycle planting management system for field crops includes the following steps:

[0126] Step 1: System Initialization and Twin Map Construction

[0127] The multi-source sensing and data acquisition module is activated, the sensor network and monitoring equipment parameters are configured, and the initial topographic data, basic soil parameters and crop variety information of the field are collected. The crop-environment digital twin modeling module builds a three-dimensional geometric model and a basic physical model of the field based on the initial data, completes the system initialization, and enters standby mode.

[0128] Step 2: Full-cycle data acquisition and preprocessing

[0129] The distributed sensing unit and the aerospace monitoring unit collect multi-source data on soil moisture, atmospheric environment, and crop physiology at a set cycle; the data preprocessing unit eliminates noise and fills in missing values; the data fusion unit realizes the spatiotemporal alignment of multi-source data, forms a standardized data set, and transmits it to the twin modeling module in real time.

[0130] Step 3: Twin Model Update and Growth Assessment

[0131] The crop-environment digital twin modeling module updates the virtual model parameters based on standardized data and verifies the consistency between the model and the physical entity through a mapping accuracy function; the full-cycle growth status assessment module calls the updated model data, calculates the growth health, classifies the growth status level, and outputs an assessment report.

[0132] Step 4: Intelligent Planting Decision Generation and Optimization

[0133] The intelligent planting decision generation module identifies the current growth stage of the crop and simulates the implementation effects of different management schemes in a virtual model. Based on the health assessment results, it generates and optimizes planting management schemes, clarifying the specific parameters and implementation requirements of irrigation, fertilization, and pesticide application measures.

[0134] Step 5: Precise Execution and Process Control

[0135] The precision execution and process control module schedules the corresponding work equipment and parses the optimized decision plan into equipment control commands. During the equipment's operation, the module monitors the work parameters in real time and adjusts the commands through a closed-loop correction mechanism to ensure that the operation is executed according to the plan.

[0136] Step 6: Dynamic Growth Prediction and Intervention

[0137] The dynamic growth prediction module predicts future crop growth trends based on current data and the trained model. If the prediction results indicate that the growth status may decline to an unfavorable level, the decision-making module is triggered to regenerate the intervention plan and take control measures in advance.

[0138] Step 7: Full-cycle data accumulation and model iteration

[0139] The twin model iteration and optimization module stores various types of data throughout the entire cycle, forming a planting data archive; it uses optimization algorithms to adjust the twin model parameters and decision logic, verifies the model performance with new data, completes model iteration, and provides an optimized foundation for the next season's planting management.

[0140] Example: Full-cycle wheat planting management scenario in the field

[0141] System Deployment:

[0142] Sensing layer: Distributed deployment of soil sensors, atmospheric sensors, and crop physiological sensors, equipped with multi-rotor drones for field remote sensing monitoring, and connected to satellite remote sensing data interface.

[0143] Computation layer: It adopts an industrial-grade server with a deep learning framework, which is responsible for twin modeling, model training and decision-making operations, and is configured with a time-series database to store planting data.

[0144] Execution layer: Deploys operational equipment such as intelligent irrigation machines, variable fertilizer machines, and plant protection drones, which are connected to the system's main control terminal via wireless communication modules.

[0145] Implementation process:

[0146] Step 1: During system initialization, collect topographic data and basic soil parameters of the wheat planting area, construct a three-dimensional geometric model and an initial physical model, and set relevant parameters for wheat varieties, such as the number of days in the growing season and photosynthetic characteristics.

[0147] Step 2: After wheat sowing, the sensor network collects data such as soil moisture, atmospheric temperature and humidity, and chlorophyll content in real time, and the drone collects canopy images weekly. The data is preprocessed and then updated to the twin model.

[0148] Step 3: During the tillering stage, the evaluation module calculated the growth health score to be 0.75, which is considered a good condition, and identified some areas with insufficient water; the twin model mapping accuracy calculation result was 0.08, which meets the accuracy requirements;

[0149] Step 4: The decision-making module simulates different irrigation schemes in the virtual model, generates precise irrigation schemes for water-scarce areas, and sets the irrigation amount and irrigation time;

[0150] Step 5: The module schedules the intelligent irrigation machine to operate according to the plan, monitors the irrigation flow in real time, and automatically adjusts the valve opening to correct the deviation when the flow deviation in a certain area exceeds the threshold.

[0151] Step 6: During the grouting period, the prediction module outputs the yield potential prediction results for the next 20 days, and generates a yield-preserving fertilization plan based on the current health assessment.

[0152] Step 7: After harvesting, the iterative module stores the full-cycle data, optimizes the photosynthetic rate parameter and evaluation index weights in the twin model, and verifies the improved accuracy of the optimized model.

[0153] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A field crop full-cycle planting management system based on digital twins, characterized by: It includes a multi-source sensing and data acquisition module, a crop-environment digital twin modeling module, a full-cycle growth status assessment module, an intelligent planting decision generation module, a precision execution and process control module, a dynamic growth prediction module, and a twin model iteration and optimization module; The multi-source sensing and data acquisition module is based on a distributed sensor network and an integrated air-space-ground monitoring device to realize the synchronous acquisition and preprocessing of field environment and crop physiological data. The crop-environment digital twin modeling module constructs a virtual mapping model of crops and planting environment, and achieves dynamic synchronization between physical entities and virtual models through a spatiotemporal alignment algorithm; The full-cycle growth status assessment module establishes a multi-dimensional assessment index system based on the output data of the twin model, and quantifies the crop growth health and potential. It includes an indicator construction unit, a data normalization unit, a health calculation unit, and a status classification unit; Indicator construction unit: Establish a three-level evaluation indicator system covering environmental suitability, physiological health, and growth consistency. Environmental suitability includes water availability and light availability; physiological health includes chlorophyll content and plant height growth rate; and growth consistency includes the coefficient of variation of plant height in the population. Data normalization unit: The min-max normalization method is used to convert indicators of different dimensions into dimensionless values ​​in the range [0,1], thereby eliminating the difference in magnitude between indicators; Health Calculation Unit: The crop growth health is calculated using a weighted summation formula, which is: Where H represents crop growth health, with a value ranging from [0,1], and a larger value indicates better growth status; n represents the number of primary indicators; p i Let α be the number of secondary indicators under the i-th primary indicator; i The weight of the i-th primary indicator satisfies β ij Let the weight of the j-th secondary indicator under the i-th primary indicator satisfy the following condition: X ij This is the normalized value of the j-th secondary indicator under the i-th primary indicator; Status grading unit: Growth status is divided into levels based on health H: H≥0.8 is excellent, 0.6≤H<0.8 is good, 0.4≤H<0.6 is average, and H<0.4 is poor. Corresponding focus points are generated for different levels. The intelligent planting decision generation module combines growth assessment results with twin simulation to generate personalized planting management plans that are adapted to the growth period. The precision execution and process control module translates decision commands into actions of the execution equipment and ensures operational accuracy through a closed-loop feedback mechanism. The dynamic growth prediction module integrates historical data and real-time parameters to establish a crop growth trend prediction model, enabling the prediction of growth period and yield potential. The twin model iteration and optimization module is based on full-cycle data accumulation and continuously optimizes the twin model parameters and decision logic through machine learning algorithms.

2. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The multi-source sensing and data acquisition module includes a distributed sensing unit, a space-air monitoring unit, a data preprocessing unit, and a data fusion unit. Distributed sensing unit: Composed of soil parameter sensors, atmospheric environment sensors and crop physiological sensors, it realizes the synchronous acquisition of data on soil moisture, atmospheric temperature and humidity and crop chlorophyll content. Space-Air Monitoring Unit: Integrates UAV remote sensing equipment and satellite remote sensing data interface to acquire macroscopic data on vegetation index and canopy temperature at the field scale; Data preprocessing unit: Kalman filtering algorithm is used to eliminate sensor noise, Lagrange interpolation is used to fill in missing data values, and time series data is smoothed. Data fusion unit: Based on DS evidence theory, it fuses multi-source data to establish a spatiotemporally consistent data set, providing input for twin modeling.

3. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The crop-environment digital twin modeling module includes a geometric modeling unit, a physical modeling unit, a spatiotemporal alignment unit, and a model accuracy verification unit; Geometric Modeling Unit: Based on UAV point cloud data and satellite imagery, construct a three-dimensional geometric model of field topography and crop canopy, preserving key features of field boundaries and irrigation facilities; Physical modeling unit: Establish crop physiological growth model and environmental action model, integrating the physiological processes of photosynthetic rate and respiration consumption with the coupling relationship of environmental factors such as light and water; Spatiotemporal alignment unit: Employs a spatiotemporal calibration algorithm to achieve temporal synchronization and spatial matching between multi-source data and virtual models, establishing a one-to-one mapping relationship between physical entities and virtual models; Model accuracy verification unit: Calculates the deviation between the virtual model and the physical entity using a mapping accuracy function, the formula of which is: Where M is the comprehensive mapping accuracy of the twin model; m is the number of parameter types involved in the verification; ω k Let be the weight coefficients of the k-th class parameter, satisfying The parameters are set according to their degree of influence on growth; V k P represents the k-th type of parameter value output by the virtual model. k The k-th type of parameter value is the actual measured value of the physical entity.

4. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The intelligent planting decision generation module includes a growth period identification unit, a simulation and deduction unit, a scheme generation unit, and a scheme optimization unit; Growth stage identification unit: Based on crop plant type parameters and phenological characteristics in the twin model, combined with the accumulated temperature model, the growth stage of the crop is automatically identified; Simulation and deduction unit: Input parameters of different management measures into the virtual model to deduce the changing trends of crop growth status under different schemes; Solution generation unit: With the goal of maximizing growth health and minimizing resource consumption, it generates an initial planting management plan, specifying the implementation time, location, and parameters of each measure; Solution optimization unit: Based on the Pareto optimality principle, the initial solution is optimized in multiple objectives, conflicting terms are eliminated, and the final executable solution is formed.

5. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The precise execution and process control module includes an equipment scheduling unit, an action analysis unit, an execution monitoring unit, and a closed-loop correction unit; Equipment scheduling unit: Based on the management plan and equipment distribution information, it uses a greedy algorithm to allocate work tasks and determine the work sequence and path of irrigation machines, fertilizer machines, drones or other equipment; Action parsing unit: converts the management parameters in the solution into control commands that the equipment can recognize; Execution monitoring unit: Collects operation parameters in real time through equipment sensors, feeds them back to the twin model, and monitors whether the operation process meets the requirements of the plan; Closed-loop correction unit: When the deviation between the operation parameters and the plan exceeds the threshold, the equipment control commands are automatically adjusted to correct the operation deviation and ensure execution accuracy.

6. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The dynamic growth prediction module includes a sample construction unit, a prediction model training unit, a prediction execution unit, and an error correction unit; Sample construction unit: Using historical planting data and real-time monitoring data as inputs, and future crop height and yield potential as outputs, a training sample set for the prediction model is constructed. Prediction model training unit: The growth prediction model is trained using a Long Short-Term Memory (LSTM) network, which integrates time series features and environmental influencing factors; Prediction Execution Unit: Inputs current data into the trained model and outputs a prediction of crop growth status over a certain future period. The prediction formula is as follows: Y t+τ =σ(W1X t +W2H t +b) Among them, Y t+τ σ is the predicted crop growth value at time τ; σ is the activation function, using the ReLU function; W1 is the input layer weight matrix; X t W1 is the input feature vector at time t, including but not limited to environmental parameters and physiological parameters; W2 is the hidden layer weight matrix; H t Let be the hidden layer state at time t; b is the bias term. Error correction unit: It adopts a rolling prediction mechanism to periodically compare the measured data with the predicted data, correct the model parameters, and reduce the prediction error.

7. The field crop full-cycle planting management system based on digital twins according to claim 1, characterized in that: The twin model iteration and optimization module includes a data storage unit, a parameter optimization unit, a logic update unit, and a model verification unit; Data storage unit: A time-series database is used to store monitoring data, execution data, prediction data and evaluation data throughout the entire cycle, and to establish a planting data archive; Parameter optimization unit: Based on the gradient descent algorithm, with the goal of minimizing the model prediction error, it continuously optimizes the physiological parameters and environmental coupling parameters in the twin model; Logical update unit: Based on new planting experience and data feedback, update the weights of growth assessment indicators and the logical parameters of decision generation rules; Model Validation Unit: The performance of the optimized model is validated using new planting cycle data to ensure improved model accuracy and adaptability.

8. The planting management method of the field crop full-cycle planting management system based on digital twins according to any one of claims 1-7, characterized in that: Includes the following steps: Step 1: System Initialization and Twin Map Construction The multi-source sensing and data acquisition module is activated, the sensor network and monitoring equipment parameters are configured, and the initial topographic data, basic soil parameters and crop variety information of the field are collected. The crop-environment digital twin modeling module builds a three-dimensional geometric model and a basic physical model of the field based on the initial data, completes the system initialization, and enters standby mode. Step 2: Full-cycle data acquisition and preprocessing The distributed sensing unit and the aerospace monitoring unit collect multi-source data on soil moisture, atmospheric environment, and crop physiology at a set cycle; the data preprocessing unit eliminates noise and fills in missing values; the data fusion unit realizes the spatiotemporal alignment of multi-source data, forms a standardized data set, and transmits it to the twin modeling module in real time. Step 3: Twin Model Update and Growth Assessment The crop-environment digital twin modeling module updates the virtual model parameters based on standardized data and verifies the consistency between the model and the physical entity through a mapping accuracy function; the full-cycle growth status assessment module calls the updated model data, calculates the growth health, classifies the growth status level, and outputs an assessment report. Step 4: Intelligent Planting Decision Generation and Optimization The intelligent planting decision generation module identifies the current growth stage of the crop and simulates the implementation effects of different management schemes in a virtual model. Based on the health assessment results, it generates and optimizes planting management schemes, clarifying the specific parameters and implementation requirements of irrigation, fertilization, and pesticide application measures. Step 5: Precise Execution and Process Control The precision execution and process control module schedules the corresponding work equipment and parses the optimized decision plan into equipment control commands. During the equipment's operation, the module monitors the work parameters in real time and adjusts the commands through a closed-loop correction mechanism to ensure that the operation is executed according to the plan. Step 6: Dynamic Growth Prediction and Intervention The dynamic growth prediction module predicts future crop growth trends based on current data and the trained model. If the prediction results indicate that the growth status may decline to an unfavorable level, the decision-making module is triggered to regenerate the intervention plan and take control measures in advance. Step 7: Full-cycle data accumulation and model iteration The twin model iteration and optimization module stores various types of data throughout the entire cycle, forming a planting data archive; it uses optimization algorithms to adjust the twin model parameters and decision logic, verifies the model performance with new data, completes model iteration, and provides an optimized foundation for the next season's planting management.

Images