Crop growth monitoring method and system based on multi-sensor fusion

Abstract

The invention relates to the field of crop monitoring and analysis, in particular to a crop growth monitoring method and system based on multi-sensor fusion. The method comprises the following steps: collecting multi-dimensional crop growth environment monitoring parameters, carrying out environment multi-dimensional perception fitting, and constructing a growth environment multi-dimensional perception model; acquiring a crop full-cycle growth monitoring image and a crop physiological state original data set; performing growth state deep evolution according to the crop physiological state original data set, and constructing a multi-mode growth state evolution graph; and carrying out adaptive region filtering and denoising on the crop full-cycle growth monitoring image flow, and carrying out crop three-dimensional morphological evolution analysis to obtain a crop three-dimensional morphological evolution rule. According to the invention, the virtual simulation model is perfected by collecting real-time parameters, intelligent analysis and real-time decision making are carried out, and the stability and accuracy of crop monitoring analysis are improved.

Description

Technical Field

[0001] The present invention relates to the field of crop monitoring and analysis, and particularly to a crop growth monitoring method and system based on multi-sensor fusion. Background Art

[0002] With the continuous development of modern agriculture towards intelligence and precision, crop growth monitoring, as a core link in agricultural production management, is gradually transforming from traditional manual observation methods to efficient and automated intelligent monitoring methods. In order to improve crop yield and quality, reduce resource waste, and achieve sustainable agricultural development, accurate and real-time monitoring of crop growth status has become a key requirement for the current development of agricultural science and technology. Especially against the background of increasingly drastic climate change and the growing shortage of agricultural labor, building an intelligent monitoring system with the help of advanced sensing technologies is of great significance for enhancing the scientific and intelligent level of agricultural production. Traditional crop monitoring methods mostly rely on manual field inspections, visual assessments, and simple instrument collections. These methods not only have low efficiency, are greatly affected by subjective factors, but also are difficult to achieve comprehensive and continuous observations of large-scale farmland, prone to problems such as data lag and incomplete information, and cannot meet the requirements of modern agriculture for high timeliness, high precision, and intelligent decision-making. Especially under variable environmental conditions, a single sensor or isolated data source often cannot fully reflect the true growth status of crops, affecting the accuracy and reliability of monitoring results.

[0003] In recent years, with the rapid development of Internet of Things technology, sensor technology, and data fusion algorithms, crop growth monitoring methods based on multi-sensor fusion have gradually become a research and application hotspot. By integrating various sensing devices such as environmental sensors (such as temperature and humidity, light, carbon dioxide concentration), image sensors, soil sensors, and meteorological sensors, all-round and multi-dimensional data collection of crop growth environment and status can be carried out, thus providing more refined and dynamic information support for agricultural management. However, in the actual application process, multi-sensor data faces many challenges such as heterogeneity, redundancy, and real-time processing. There are significant differences in the data formats, update frequencies, and accuracies collected by different sensors. How to effectively fuse, intelligently analyze, and make real-time decisions on massive and multi-source monitoring data has become a bottleneck problem in the current research of agricultural informatization technology. In addition, factors such as environmental interference, sensor failures, and data anomalies may also affect the stability and accuracy of the monitoring system. Therefore, an intelligent crop growth monitoring method is needed. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a crop growth monitoring method and system based on multi-sensor fusion to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a method for monitoring crop growth based on multi-sensor fusion, comprising the following steps: Step S1: Collect multi-dimensional monitoring parameters of the crop growth environment, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional perception model of the growth environment; Step S2: Obtain the full-cycle growth monitoring images of the crops and the original dataset of the crop physiological states; perform deep evolution of the growth states based on the original dataset of the crop physiological states, and construct a multi-modal growth state evolution map; Step S3: Perform adaptive region filtering and denoising on the full-cycle growth monitoring image stream of the crops, and perform three-dimensional morphological evolution analysis of the crops to obtain the three-dimensional morphological evolution law of the crops; Step S4: Perform dynamic growth behavior change analysis on the three-dimensional morphological evolution law of the crops according to the multi-modal growth state evolution map, and perform digital simulation of the growth trend according to the multi-dimensional perception model of the growth environment, and construct a twin model of the crop growth trend; Step S5: Perform multi-time scale growth simulation on the twin model of the crop growth trend, and perform deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; Step S6: Perform growth state simulation drive on the virtual-real resonance growth trend evolution model, and perform optimal intervention and regulation decision-making to construct an intelligent monitoring and regulation strategy for crops.

[0006] In this specification, a crop growth monitoring system based on multi-sensor fusion is provided for executing the method for monitoring crop growth based on multi-sensor fusion as described above, comprising: An environmental perception module for collecting multi-dimensional monitoring parameters of the crop growth environment, performing environmental multi-dimensional perception fitting, and constructing a multi-dimensional perception model of the growth environment; A growth state evolution module for obtaining the full-cycle growth monitoring images of the crops and the original dataset of the crop physiological states; performing deep evolution of the growth states based on the original dataset of the crop physiological states, and constructing a multi-modal growth state evolution map; A morphological evolution module for performing adaptive region filtering and denoising on the full-cycle growth monitoring image stream of the crops, and performing three-dimensional morphological evolution analysis of the crops to obtain the three-dimensional morphological evolution law of the crops; A trend simulation module for performing dynamic growth behavior change analysis on the three-dimensional morphological evolution law of the crops according to the multi-modal growth state evolution map, and performing digital simulation of the growth trend according to the multi-dimensional perception model of the growth environment, and constructing a twin model of the crop growth trend; A deviation compensation module for performing multi-time scale growth simulation on the twin model of the crop growth trend, and performing deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; An intervention and regulation module is used to perform growth state simulation driving on the virtual-real resonance growth trend evolution model, make optimal intervention and regulation decisions, and construct an intelligent monitoring and regulation strategy for crops.

[0007] The beneficial effects of the present invention are specifically as follows: By collecting multi-dimensional environmental data such as temperature, humidity, light, soil pH, water content, CO2 concentration, etc., the high-precision restoration of the crop growth environment is realized. Through non-linear multiple fitting, the comprehensive influence relationship of different environmental factors on the crop growth state is revealed. A quantifiable and computable environmental input variable framework is provided for subsequent growth state simulation and regulation. Combining multi-time period image data with original physiological indicators (such as chlorophyll concentration, transpiration rate, photosynthesis efficiency, etc.), the life process of the crop is comprehensively recorded. Through deep learning technology, the image features and physiological states are fused to reveal the evolution law from visual morphology to internal physiology. A time-continuous growth state map is constructed for each type of crop to support subsequent behavior prediction and trend deduction. Adaptive regional filtering can effectively remove interference noises such as wind blowing, water mist, and light spots, improving the stability of the monitoring images. A three-dimensional model is constructed based on multi-angle images to analyze the dynamic change characteristics such as crop height, crown width, and leaf expansion. The growth rhythm and morphological expansion trajectory of different parts during the crop development process are established, providing a spatial feature basis for model prediction. Through the fusion of the map and three-dimensional morphological data, a typical growth behavior model of the crop at different stages is established. Using the multi-dimensional environmental perception model and the three-dimensional change law of the crop to jointly drive digital simulation, complex state simulation is realized. This model can simulate the real growth state changes in a virtual environment, improving the prediction accuracy of the crop response under different environmental conditions. Multi-time scale modeling enables the system to adjust in real time and plan future growth strategies. Comparing the simulation results with the actual growth state forms a feedback loop, continuously optimizing the simulation parameters and enhancing the authenticity of the twin model. A closed-loop driving system is established between the simulation and the actual situation to ensure that the simulation system not only "resembles reality" but also can affect real management. Different fertilization, irrigation, shading, and ventilation strategies are tested in a virtual simulation environment to find the optimal combination and reduce the field trial and error cost. By backtracking from the simulation results, the most suitable intervention timing, dosage, and frequency are determined to improve the resource utilization efficiency. An automatic perception - simulation evolution - intelligent regulation closed-loop system is realized, promoting the development of crops towards intelligence, adaptability, and precise scheduling. Brief Description of the Drawings

[0008] Figure 1 It is a schematic flow chart of the steps of a method for monitoring crop growth based on multi-sensor fusion according to the present invention; Figure 2 It is a schematic detailed implementation step flow chart of step S1; Figure 3 It is a schematic detailed implementation step flow chart of step S2; Figure 4It is a schematic diagram of the detailed implementation steps of step S3. Specific implementation manner

[0009] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0010] The embodiments of the present application provide a method and system for monitoring crop growth based on multi-sensor fusion. The execution subjects of the method and system for monitoring crop growth based on multi-sensor fusion include, but are not limited to, mechanical equipment, data processing platforms, cloud server nodes, network upload devices, etc. that carry the system, which can be regarded as general computing nodes of the present application. The data processing platform includes, but is not limited to, at least one of an audio and image management system, an information management system, and a cloud data management system.

[0011] Please refer to Figures 1 to 4 , the present invention provides a method for monitoring crop growth based on multi-sensor fusion. The method for monitoring crop growth based on multi-sensor fusion includes the following steps: Step S1: Collect multi-dimensional crop growth environment monitoring parameters, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional perception model of the growth environment; Step S2: Obtain the full-cycle growth monitoring images of the crops and the original dataset of the crop physiological states; perform deep evolution of the growth states according to the original dataset of the crop physiological states, and construct a multi-modal growth state evolution map; Step S3: Perform adaptive region filtering and denoising on the full-cycle growth monitoring image stream of the crops, and perform three-dimensional morphological evolution analysis of the crops to obtain the three-dimensional morphological evolution law of the crops; Step S4: Perform dynamic growth behavior change analysis on the three-dimensional morphological evolution law of the crops according to the multi-modal growth state evolution map, and perform digital simulation of the growth trend according to the multi-dimensional perception model of the growth environment to construct a crop growth trend twin model; Step S5: Perform multi-time scale growth simulation on the crop growth trend twin model, and perform deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; Step S6: Perform growth state simulation drive on the virtual-real resonance growth trend evolution model, and perform optimal intervention and regulation decision-making to construct a crop intelligent monitoring and regulation strategy.

[0012] In the embodiments of the present invention, refer to Figure 1 , it is a schematic diagram of the step flow of a method for monitoring crop growth based on multi-sensor fusion of the present invention. In this example, the steps of the method for monitoring crop growth based on multi-sensor fusion include: Step S1: Collect multi-dimensional monitoring parameters of the crop growth environment, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional perception model of the growth environment; In this embodiment, a variety of sensors are reasonably arranged in the farmland, including soil moisture sensors, pH sensors, conductivity sensors, and weather stations (for measuring temperature, humidity, and light intensity). Each sensor node should cover a specific area to ensure that the collected environmental data is representative. Each sensor collects environmental monitoring data in real time according to the set sampling frequency (e.g., 1Hz). Assume that the soil moisture sensor collects the humidity value once per minute within a range of 10 square meters, and the weather station records the temperature and humidity, etc. once per minute. The collected data is transmitted to the central database through wireless communication technologies (such as LoRa or Zigbee). Set the data storage format as CSV for subsequent data processing and analysis. During the data collection process, regularly check the working status of the sensors to ensure the accuracy of the data. If abnormal data (such as a humidity value exceeding 100%) is found, the system will automatically mark and record it for subsequent processing. Extract the required environmental monitoring parameters from the database, including soil moisture, pH value, conductivity, temperature, humidity, and light intensity. Clean the data to remove missing values and outliers. Use interpolation methods (such as linear interpolation) to fill in the missing values to ensure the continuity of the data. According to the needs of crop growth, select key environmental monitoring parameters as the input features of the model. Assume that soil moisture, temperature, and light intensity are selected as the model inputs, and pH value and conductivity are used as auxiliary features. Select a suitable machine learning algorithm (such as linear regression, decision tree, or random forest) for model training. Divide the dataset into a training set (70%) and a test set (30%), use the training set to train the model, and ensure the generalization ability of the model through cross-validation. Use the test set to evaluate the performance of the model and calculate metrics such as RMSE. If the RMSE of the model is 2.5%, it indicates that the model has a good fitting effect on the environmental parameters. Adjust the model parameters according to the evaluation results to optimize the performance. Apply the constructed multi-dimensional perception model to real-time data analysis to monitor and predict the crop growth environment in real time. Continuously update and optimize the model to ensure its adaptation to environmental changes.

[0013] Step S2: Obtain the full-cycle growth monitoring images of the crops and the original dataset of the crop physiological states; perform deep evolution of the growth states based on the original dataset of the crop physiological states, and construct a multi-modal growth state evolution map; In this embodiment, multiple sensor devices are used to obtain the monitoring images of the whole growth cycle of crops. The goal of this step is to ensure that clear image data can be captured at different stages of crop growth for subsequent analysis. Set the experimental parameters, for example, select the resolution of image acquisition as 1920x1080 (1080p), and set the acquisition frequency as once a day to cover the entire growth cycle. During the process of obtaining images, an unmanned aerial vehicle (UAV) equipped with a high-resolution camera can be used for aerial photography, or multiple fixed cameras can be deployed on the ground to monitor specific areas. The UAV can cover a large area of farmland and take images at different heights and angles to ensure that the obtained images have a good perspective and coverage. When collecting images, attention should be paid to the change of lighting conditions, so multiple shots are taken on sunny and cloudy days to ensure the diversity and reliability of the data. All the obtained image data will be uploaded to the central data storage system for unified management and processing. During the data storage process, each image needs to be marked with a timestamp, location, and crop type for subsequent analysis and traceability. This process ensures that the AGV system can efficiently and accurately capture the growth status of crops and provides basic data for subsequent physiological state analysis. Obtain the original dataset of the physiological state of crops. The goal of this step is to collect the physiological data related to crop growth for comprehensive growth status analysis. Set the experimental parameters, for example, select the measurement frequency as once a week to capture the changes in the physiological state. The physiological state data can be collected through sensors (such as soil moisture sensors, temperature sensors, and light sensors), which can monitor key indicators such as the soil moisture content, temperature change, and light intensity in real time to ensure that the AGV can obtain the environmental parameters of crop growth in a timely manner. At the same time, the plant physiological indicators, including the chlorophyll content of leaves, leaf area index (LAI), and plant height, can be measured regularly manually or automatically. All the physiological state data will be stored together with the image data and integrated through the data management system. Each set of physiological data should include a timestamp, measurement location, and relevant meteorological information (such as precipitation and wind speed) for subsequent analysis and comparison. This process ensures the comprehensive monitoring of the crop growth process and provides the necessary data support for the deep evolution of the growth state. After obtaining the monitoring images of the whole growth cycle and the original dataset of the physiological state, conduct the in-depth evolution analysis of the growth state. The goal of this step is to combine the physiological state and image data to extract the key evolution characteristics in the crop growth process. Set the experimental parameters, for example, set the time window for data fusion as once a week to observe the dynamic changes in the growth state.

[0014] Through data fusion technology, the monitoring images and physiological state data are integrated. The image data is processed, including denoising, enhancing contrast, and edge detection, to improve the image quality. Then, computer vision algorithms (such as the deep learning model Convolutional Neural Network CNN) are applied to extract features from the processed images, identifying the growth characteristics of crops and signs of pests and diseases. Machine learning algorithms (such as random forest or support vector machine) are used to perform correlation analysis between the extracted features and the physiological state data. By constructing a regression model, the relationship between physiological indicators (such as leaf area index, soil moisture, etc.) and the growth state is analyzed, thereby identifying the key factors affecting crop growth. This step will generate a comprehensive set of growth state evolution indicators that can reflect the health status and growth trends of crops at different growth stages, providing a basis for subsequent optimization management. According to the results of the in-depth evolution analysis of the growth state, a multi-modal growth state evolution map is constructed. The goal of this step is to present various physiological states and image data in a visual manner for agricultural managers to make decisions. Experimental parameters are set, such as setting the map update frequency to once a month to ensure the display of the latest growth state. The analysis results are visualized through data visualization tools (such as Tableau or Matplotlib in Python). The growth state evolution indicators are presented in the form of a time series to show the growth trends of crops at different time points. At the same time, heat maps or scatter plots are used to show the relationship between different physiological indicators and the crop growth state, such as the relationship between soil moisture and leaf area index.

[0015] Step S3: Perform adaptive regional filtering denoising on the full-cycle growth monitoring image stream of crops and conduct three-dimensional morphological evolution analysis of crops to obtain the three-dimensional morphological evolution law of crops; In this embodiment, adaptive region filtering denoising is performed on the acquired crop full-cycle growth monitoring image stream. The goal of this step is to remove the noise in the image, improve the image quality, and provide clear basic data for subsequent three-dimensional morphology analysis. Set experimental parameters, for example, select the filtering window size to be 5x5 pixels to balance the denoising effect and detail retention. The process of adaptive region filtering includes the following steps. The image is initially processed using Gaussian filtering or median filtering. Gaussian filtering can effectively reduce high-frequency noise by performing weighted averaging on the area around each pixel. Median filtering, on the other hand, replaces the value of each pixel with the median of its neighboring pixels and is particularly suitable for removing salt-and-pepper noise. An adaptive filtering method is adopted to dynamically adjust the filtering parameters according to the local characteristics of the image. Specifically, the adaptive mean filtering algorithm can be used. This algorithm calculates the mean and variance of the local area of the image and dynamically adjusts the weights of the filter. By setting a threshold, it is determined whether a certain area needs to be denoised. If the variance of the area exceeds the threshold, the filtering effect is enhanced. When processing the image, the denoising effect should be evaluated regularly, and the signal-to-noise ratio (SNR) is calculated to quantify the denoising effect. Set the target SNR to be above 30 dB to ensure that the image quality meets the requirements of subsequent analysis. The image processed by adaptive region filtering will be used for subsequent three-dimensional morphology evolution analysis to ensure the accuracy and reliability of the analysis results. After the denoising process is completed, three-dimensional morphology reconstruction of the crop is performed. The goal of this step is to convert the two-dimensional monitoring image into a three-dimensional model for in-depth analysis of the morphological evolution law of the crop. Set experimental parameters, for example, select the point cloud density to be 1000 points per square meter to ensure the detail performance of the three-dimensional model. Use structured light or stereo vision technology to capture the three-dimensional information of the crop through multi-view images. Use algorithms (such as SIFT or ORB) to extract feature points and calculate the corresponding relationships between feature points under different perspectives through a matching algorithm, thereby generating dense point cloud data. This process can be achieved by using multiple cameras to simultaneously capture images from different angles.

[0016] Through 3D reconstruction algorithms (such as Poisson reconstruction or triangular mesh generation), the generated point cloud data is converted into a 3D model. The Poisson reconstruction algorithm can effectively handle sparse and noisy data and generate a smooth surface. During the process of generating the 3D model, smoothing parameters are set to control the smoothness of the model and ensure the balance between details and overall effects. After completing the 3D model, visualization tools (such as Meshlab or Blender) are used to render the model to generate an interactive 3D view, which is convenient for subsequent analysis and display. This process ensures that the 3D information of the crop morphology is completely captured, providing basic data for the analysis of morphological evolution. After generating the 3D model of the crop, 3D morphological evolution analysis is carried out. The goal of this step is to study the morphological changes of the crop during the growth process and identify the growth rules and potential problems. Experimental parameters are set, for example, the analysis time window is set to once a week to observe the dynamic characteristics of morphological changes. Through time series analysis of the 3D model, the 3D morphologies at different growth stages are compared, and key morphological features are extracted, such as leaf area, stem height, and the number of branches. Using morphological analysis methods, morphological indexes at different time points are calculated. The leaf area can be calculated by integrating the surface of the 3D model to obtain the photosynthetically active area of the crop at different growth stages. To gain a deep understanding of the morphological evolution rules, statistical analysis methods (such as analysis of variance or multiple regression analysis) can be used to explore the relationship between environmental factors (such as light, temperature, humidity, etc.) affecting crop growth and crop morphological changes. By establishing a model, the correlation between the morphological features at key growth stages and environmental variables is identified.

[0017] Step S4: Based on the multi-modal growth state evolution atlas, conduct dynamic growth behavior change analysis on the 3D morphological evolution law of crops, and perform digital simulation of the growth trend according to the multi-dimensional perception model of the growth environment to construct a twin model of the crop growth trend; In this embodiment, based on the multi-modal growth state evolution map, the dynamic growth behavior changes of the three-dimensional morphological evolution law of crops are analyzed. The goal of this step is to identify and understand the dynamic change characteristics of crops during the growth process and explore the relationship between the growth state and environmental factors. Set experimental parameters, such as setting the analysis time window to once a week to capture the changes in the growth state. When conducting dynamic growth behavior analysis, it is first necessary to deeply analyze the multi-modal growth state evolution map. The map contains the physiological indicators, environmental parameters, and three-dimensional morphological characteristics of crops at different growth stages. Using data mining techniques (such as clustering analysis or association rule mining), identify the key behavior patterns at different growth stages. The physiological indicators (such as leaf area index, soil moisture, etc.) at different growth stages can be classified into several categories through a clustering algorithm to determine the growth characteristics of each stage. Apply time series analysis methods to monitor the change trend of specific physiological indicators over time. The ARIMA model (Autoregressive Integrated Moving Average model) can be used to predict the future growth state and evaluate the accuracy and reliability of the model. During the analysis process, pay attention to the main environmental factors affecting growth (such as light, temperature, humidity, etc.), and explore the relationship between these factors and the growth state of crops through regression analysis. Form a set of dynamic growth behavior change analysis reports to identify the key growth stages and behavior change characteristics. This process not only helps to understand the growth law of crops but also provides data support for subsequent digital simulations. After completing the dynamic growth behavior change analysis, construct a multi-dimensional perception model of the growth environment. The goal of this step is to comprehensively capture the environmental factors affecting crop growth for more accurate growth trend simulation. Set experimental parameters, such as setting the sampling frequency of environmental variables to once an hour to ensure the real-time nature of the data.

[0018] The construction of the multi-dimensional perception model involves the monitoring of multiple key environmental variables, such as soil moisture, temperature, light intensity, CO2 concentration, etc. Through the sensor network, these data are collected and stored in real time. Data fusion algorithms (such as Kalman filtering or particle filtering) are used to process the sensor data to eliminate noise and improve the accuracy of the data. Machine learning methods (such as random forest or support vector machine) are applied to establish a relationship model between environmental variables and the growth status of crops. By training on historical data, the degree of influence of environmental factors on the growth status is identified. Experimental parameters are set, such as selecting 80% of the data for training and 20% of the data for validation, to ensure the generalization ability of the model. The constructed multi-dimensional perception model will provide basic data for the digital simulation of the growth trend of crops, ensuring the accuracy and reliability of the simulation. According to the constructed multi-dimensional perception model of the growth environment, the digital simulation of the growth trend of crops is carried out. The goal of this step is to present the growth dynamics of crops and their relationship with the environment in digital form. Experimental parameters are set, such as setting the simulation time span to 6 months to cover the entire growth cycle of crops. When conducting digital simulation, initial conditions need to be set first, including the type of crops, initial growth status, and environmental parameters. Based on the multi-dimensional perception model, simulation algorithms (such as discrete event simulation or system dynamics model) are used to simulate the growth process of crops under different environmental conditions. By continuously updating environmental parameters, the changes of crops at different growth stages are simulated. The Monte Carlo simulation method can be used to generate multiple possible environmental states through random sampling, and multiple simulations are carried out according to these states to obtain the growth trend of crops. This method can reflect the impact of environmental changes on crop growth and make the simulation results more reliable. The generated digital simulation results of the growth trend will provide intuitive decision-making support for crop management, helping farmers predict future growth status and formulate corresponding management strategies.

[0019] According to the results of digital simulation of the growth trend, a twin model of the growth trend of crops is constructed. The goal of this step is to achieve real-time synchronization between virtual crops and the growth of actual crops, providing data support for intelligent agricultural management. Set experimental parameters, such as setting the model update frequency to once a week, to ensure matching with the actual growth state. The construction of the twin model includes comparing and calibrating the digital simulation results with actual monitoring data. Using data fusion technology, data from different sources (such as sensor data, image data, simulation results) are integrated to form a comprehensive growth trend model. By monitoring the growth state of crops in real time, the parameters of the twin model are adjusted in a timely manner to ensure its consistency with the actual crop state. In this process, a feedback control mechanism is applied to monitor the deviation between the actual growth state and the simulation results, and the model is continuously optimized based on these deviations. If it is found that the actual leaf area index is lower than the simulated value, the simulation parameters are adjusted to better reflect the actual situation. The constructed twin model of the growth trend of crops will provide real-time data support for agricultural management, helping farmers optimize management decisions and improve agricultural production efficiency. This process not only improves the accuracy of crop growth monitoring but also lays a foundation for promoting the development of intelligent agriculture.

[0020] Step S5: Conduct multi-time scale growth simulations on the twin model of the growth trend of crops, and perform deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; In this embodiment, multi-time scale growth simulations are conducted on the twin model of the growth trend of crops. The goal of this step is to deeply analyze the growth dynamics of crops at different time scales and explore the potential change laws during the growth process. Set experimental parameters, such as setting the simulation time span to 12 months and dividing the time scale into three levels: daily, weekly, and monthly, in order to comprehensively capture the subtle changes in the growth process. When conducting multi-time scale growth simulations, it is first necessary to select an appropriate growth model. Usually, physiology-based growth models (such as DSSAT or APSIM) are adopted. These models can comprehensively consider the influence of environmental factors such as light, temperature, water, and nutrients on crop growth. By inputting the current environmental parameters and the initial growth state, the model will predict the growth of crops at different time scales. During the simulation process, by setting different environmental conditions (such as light intensity, precipitation, soil moisture, etc.), observe the responses of crops at different growth stages. On the daily scale, the impact of daily light changes on crop photosynthesis can be simulated; on the weekly scale, observe how the weekly temperature changes affect the growth rate; on the monthly scale, analyze the overall impact of long-term drought on crop growth.

[0021] The generated multi-time scale growth simulation results will form a dynamic growth trend map, showing the growth trends and change characteristics of crops at different time scales. This process will provide basic data for subsequent deviation feedback compensation, enabling the model to better match the actual growth situation. After completing the multi-time scale growth simulation, deviation feedback compensation is required to ensure the consistency between the simulation results and the actual crop growth state. The goal of this step is to adjust the parameters of the growth model through real-time monitoring and feedback mechanisms to improve the accuracy of the model. Set experimental parameters, such as setting the feedback update frequency to once a week, to ensure timely response to environmental changes. The process of deviation feedback compensation includes the following steps. Real-time collect the growth monitoring data of crops, including physiological indicators and environmental parameters, and obtain the latest growth status information through the sensor network. These data will be compared with the results of the multi-time scale growth simulation to identify the sources of deviation. By calculating the deviation value, for example: Deviation = Actual growth state - Simulated growth state. If it is found that the actual growth state is lower than the simulated value, the system will automatically trigger the compensation mechanism to adjust the parameters of the model. An adaptive control algorithm (such as PID control) can be used to achieve this process. Set appropriate proportional, integral, and differential parameters so that the deviation can be quickly corrected. If the leaf area index predicted by the model is higher than the actual value, the system will increase the weights of soil moisture or light intensity to better reflect the actual growth situation. By continuously iterating this process, the model can adapt to environmental changes in real time and maintain a high prediction accuracy.

[0022] Based on the results of deviation feedback compensation, construct a virtual-real resonance growth trend evolution model. The goal of this step is to achieve a high degree of consistency between the virtual model and the actual growth state, providing real-time and accurate data support for crop management. Set experimental parameters, such as setting the accuracy of model adjustment to within 5%, to ensure the reliability of the simulation results. The construction of the virtual-real resonance growth trend evolution model includes combining real-time data with the results of multi-time scale growth simulation. Through data fusion technology, compare the actual monitoring data with the simulation results to form a comprehensive growth trend model. Use machine learning methods (such as deep learning) to identify the relationship between the virtual and the actual by training the model and optimize the parameters of the model. In this process, a recurrent neural network (RNN) or a long short-term memory network (LSTM) can be used to capture the dynamic change characteristics in time series data. These algorithms can effectively process time series data, learn the potential patterns in the growth trend, and thus improve the prediction ability of the model.

[0023] Step S6: Drive the growth state simulation of the virtual-real resonance growth trend evolution model and make an optimal intervention and regulation decision to construct an intelligent monitoring and regulation strategy for crops.

[0024] In this embodiment, a growth state simulation drive is performed on the virtual-real resonance growth trend evolution model. The goal of this step is to predict and analyze the growth state of crops in real time through simulation technology to assist in formulating corresponding management strategies. Set experimental parameters, such as a simulation time span of 6 months and a simulation frequency of once a week to ensure timely response to growth dynamics. The process of growth state simulation drive first requires input of current environmental parameters and crop growth states. Use the established virtual-real resonance model to combine actual monitoring data with simulation data. Set initial conditions according to the growth requirements of crops, such as soil humidity, nutrient levels, and current climate conditions. During the simulation process, use model-based prediction algorithms (such as system dynamics models) to perform dynamic simulation of the growth state. These models can comprehensively consider the effects of various environmental factors (such as temperature, humidity, light, etc.) on crop growth. By inputting real-time environmental data, the model will be updated in real time and predict the growth trend and state changes of the crops. If the current soil humidity is 30%, the model will simulate the effects of different humidity levels on crop growth and predict the growth state in the next few weeks. By adjusting different environmental variables and observing their effects on crop growth, the simulation results will provide data support for subsequent intervention and regulation decisions. After constructing a multi-scenario perturbation simulation matrix, perform a growth state simulation drive on the virtual-real resonance growth trend evolution model based on this matrix. This step aims to evaluate the growth state of crops through different environmental scenarios. Input the temperature, humidity, and light parameters of each scenario into the virtual-real resonance growth trend evolution model. The model can adopt a dynamic system model, where environmental variables affect the growth rate and growth form of crops.

[0025] Assume the inputs to the model include: Input for Scenario 1: Temperature 20°C, humidity 50%, light 600 µmol / m² / s Input for Scenario 2: Temperature 30°C, humidity 70%, light 1000 µmol / m² / s Then, calculate the growth states under different scenarios through numerical simulation methods (such as the finite difference method or Monte Carlo simulation). The model predicts that in Scenario 1, the height of the crop increases by 3 cm, while in Scenario 2, the height increases by 5 cm. In this way, generate simulation data of the growth state under multiple perturbations, and record parameters such as the height, leaf area, and biomass of the crop corresponding to each scenario. These data will provide a basis for subsequent risk identification and state assessment. After obtaining the simulation data of the growth state, conduct potential growth risk identification. This step aims to identify potential risk points by analyzing the growth state data in order to take corresponding intervention measures. Set the criteria for risk identification. Assume that defining a growth rate lower than the expected value (such as 1 cm / week) or a decrease in biomass (such as lower than 100 g / m²) as potential risks. Through statistical analysis of the simulation data, check whether the growth state in each scenario meets these criteria. Assume that in Scenario 3, the predicted growth rate by the model is 0.8 cm / week, and the biomass is 90 g / m², indicating the existence of potential growth risks in this scenario. Record all the identified risk points and analyze their possible causes (such as high temperature, low humidity, etc.). Through these analysis results, it can provide data support for subsequent state assessment and lay a foundation for intervention decisions. After potential growth risk identification, extract multiple parameters based on the simulation data of the growth state and conduct a comprehensive growth state assessment to generate a simulation assessment report of the growth state. Extract key parameters from the simulation data, such as growth height, leaf area, photosynthesis efficiency, and biomass. Use the weighted average method to calculate the comprehensive assessment score according to the importance of different parameters. Assume that the growth height accounts for 40%, the leaf area accounts for 30%, and the biomass accounts for 30%, and a comprehensive score can be calculated. Use statistical analysis methods (such as principal component analysis PCA) to analyze the extracted parameters and identify the main factors affecting the growth state. Assume that the analysis results show that temperature and light have the greatest impact on growth. Based on the assessment results, generate a simulation assessment report of the growth state, which includes the growth state scores of each scenario, potential risk points and their cause analysis. This report will provide a scientific basis for subsequent intervention and regulation decisions. According to the potential growth state risk points and the simulation assessment report of the growth state, make an optimal intervention and regulation decision and construct an intelligent monitoring and regulation strategy for crops. Analyze the risk points identified in the assessment report and determine the intervention measures to be taken. If the risk in a certain scenario is due to insufficient light, then increasing the light intensity or adjusting the planting density can be considered. Combine real-time monitoring data and use a decision support system (DSS) to optimize the intervention strategy. By simulating the effects of different intervention schemes (such as increasing water or fertilization), select the optimal scheme. Assume that after simulation, it is decided to reduce the temperature and increase the fertilization amount in the high-temperature scenario. Construct an intelligent monitoring and regulation strategy for crops to ensure that management measures can be flexibly adjusted in real-time monitoring. By implementing these strategies, the growth risks can be effectively reduced and the growth efficiency of crops can be improved.

[0026] In this embodiment, referring to Figure 2 , which is a schematic diagram of the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Collecting multi-dimensional crop growth environment monitoring parameters based on multi-sensor nodes; Performing spatio-temporal standard optimization on the crop growth environment monitoring parameters to construct repositioned standard environment monitoring parameters, where the repositioned standard environment monitoring parameters include multi-dimensional soil monitoring parameters, atmospheric environment temperature and humidity parameters, and environmental real-time light intensity; Calculating the temporal fluctuations of the moisture content, conductivity, and pH value of the multi-dimensional soil monitoring parameters, and extracting soil temporal fluctuation characteristics; Mining the temperature and humidity spatial distribution changes of the atmospheric environment temperature and humidity parameters to construct an environmental temperature and humidity spatial distribution change map; Analyzing the light intensity change trend of the environmental real-time light intensity to obtain the environmental light intensity change trend; Performing environmental multi-dimensional perception fitting based on the soil temporal fluctuation characteristics, environmental light intensity change trend, and environmental temperature and humidity spatial distribution change map to construct a growth environment multi-dimensional perception model.

[0027] In this embodiment, a multi-sensor node network is established to collect multi-dimensional monitoring parameters of the crop growth environment in real time. The sensor nodes include a soil moisture sensor, a soil conductivity sensor, a pH sensor, an air temperature and humidity sensor, and a light intensity sensor. These sensors are distributed at different locations in the farmland to ensure the representativeness and comprehensiveness of the data. In the experimental design, it is assumed that each sensor node collects data every 5 minutes. The soil moisture sensor will obtain the water content of the soil, the conductivity sensor will provide the soil conductivity data, and the pH sensor is used to measure the acidity and alkalinity of the soil. At the same time, the air temperature and humidity sensor records the air temperature and humidity, and the light intensity sensor measures the light intensity of the environment. In a 10-acre farmland, 10 sensor nodes are set, and each node is responsible for monitoring a specific area. In this way, after a period of operation, rich monitoring data can be obtained, including parameters in multiple dimensions such as soil water content, pH value, conductivity, environmental temperature, humidity, and light intensity. After the data collection is completed, spatio-temporal standard optimization is carried out to construct repositioning standard environmental monitoring parameters. This step involves cleaning, normalizing, and standardizing the collected monitoring data to ensure the accuracy and consistency of the data. Missing value processing and outlier detection are performed on the data collected by each sensor. The mean imputation method can be used to fill in the missing values, and box plots are used to detect and remove outliers. Then, the data is sorted according to the time series to ensure the timeliness of the data. Repositioning standard environmental monitoring parameters are constructed. It is assumed that the repositioning standard parameters include the water content of the soil (range 0-100%), conductivity (range 0-5 mS / cm), pH value (range 4-8), and environmental temperature and humidity (temperature range -10°C to 40°C, humidity range 0-100%) and light intensity (range 0-2000 µmol / m² / s). By standardizing the data at each monitoring point, data from different sources are unified into a standard range for subsequent analysis. Time series fluctuation analysis is performed on the water content, conductivity, and pH value of the multi-dimensional soil monitoring parameters to extract soil time series fluctuation characteristics. This process can be achieved through time series analysis methods. Statistical methods are used to perform time series analysis on the water content, conductivity, and pH value of the soil. By calculating the moving average and volatility, the trend of their change over time is revealed. A 7-day sliding window is set, and the average value and standard deviation of each parameter within this window are calculated to identify the fluctuation of the parameter. Secondly, Fourier transform is used to analyze the spectral characteristics to identify periodic fluctuations. It is assumed that the soil water content monitored within a month shows periodic changes with a frequency of once a week. This analysis can help farmers understand the change law of soil moisture and provide a basis for reasonable irrigation.

[0028] Mining the spatial distribution changes of atmospheric environmental temperature and humidity parameters to construct a spatial distribution change map of environmental temperature and humidity. This step aims to analyze the temperature and humidity characteristics in different regions and their changes over time. Visualize the temperature and humidity data through a Geographic Information System (GIS). Map the temperature and humidity data of each sensor node to the geographical location of the farmland to construct a spatial distribution map of temperature and humidity. Use an interpolation method (such as Kriging interpolation) to generate a temperature and humidity distribution map for the entire farmland. Then, analyze the data changes at different time points to construct a temperature and humidity change map. By comparing the distribution maps of different time periods, identify the change trends of temperature and humidity. Assume that during the growing season, the temperature in some areas is significantly higher than that in other areas, indicating possible growth problems caused by uneven temperature. Analyze the change trend of real-time environmental light intensity to understand the impact of light intensity on crop growth. Extract the change trend of light intensity through data processing and analysis. Collect light intensity data from light intensity sensors at different time periods for time series analysis. Use linear regression or moving average method to analyze the change trend of light intensity. Assume that the light intensity shows a gradually increasing trend during a growth cycle, and record the time points of the peak and valley values. Secondly, by analyzing the relationship between light intensity and other environmental parameters (such as temperature and humidity, soil moisture content), identify its impact on crop growth. Assume that there is a positive correlation between light intensity and soil moisture content, indicating that when the light intensity increases, the evaporation rate of soil moisture accelerates.

[0029] Based on the temporal fluctuation characteristics of the soil, the change trend of environmental light intensity, and the spatial distribution change map of environmental temperature and humidity, construct a multi-dimensional perception model of the growth environment. This model can comprehensively consider the impact of multiple parameters on crop growth, thereby providing more comprehensive environmental monitoring and analysis. Use machine learning algorithms (such as random forest or support vector machine) to train the collected multi-dimensional data to construct a prediction model. The model inputs are parameters such as soil moisture content, conductivity, pH value, light intensity, environmental temperature and humidity, etc., and the output is the prediction result of the crop growth state. During the training process, evaluate the accuracy of the model through cross-validation and continuously optimize the hyperparameters. Assume that after training, the prediction accuracy of the model reaches more than 90%. The constructed multi-dimensional perception model can monitor and analyze environmental changes in real time, providing a scientific basis for crop growth management.

[0030] In this embodiment, the specific steps for optimizing the spatio-temporal standard of the crop growth environment monitoring parameters and constructing the repositioned standard environmental monitoring parameters are as follows: Calculate the sampling frequency deviation between devices for multiple sensor nodes and extract the sampling frequency deviation of multiple sensors; Calibrate the timestamps of the growth environment monitoring parameters according to the sampling frequency deviation of multiple sensors to obtain unified time-series monitoring parameters; Calculate the spatial positions of multiple sensor nodes and extract the position coordinates of each node; Construct a spatial node coordinate network based on the position coordinates of each node; Perform multi-parameter spatial repositioning on the unified time-series monitoring parameters based on the spatial node coordinate network to obtain the repositioned standard environmental monitoring parameters.

[0031] In this embodiment, deviation calculations are performed on the sampling frequencies of multiple sensor nodes. This step aims to identify and quantify the sampling frequency inconsistencies between different sensors to ensure the accuracy of subsequent data processing. Suppose there are five sensor nodes (A, B, C, D, E) in a monitoring network, and the ideal sampling frequency of each node is set to 1 Hz. By recording the actual number of samples taken by each node within a specific time period, the actual sampling frequency is calculated. Within 10 minutes, node A sampled 58 times, node B sampled 62 times, node C sampled 55 times, node D sampled 60 times, and node E sampled 61 times. By calculating the actual sampling frequency of each node, we get: Node A: 58 / 600 = 0.097 Hz Node B: 62 / 600 = 0.103 Hz Node C: 55 / 600 = 0.092 Hz Node D: 60 / 600 = 0.1 Hz Node E: 61 / 600 = 0.102 Hz Calculate the sampling frequency deviation of each node: Deviation A = 0.097 - 1.0 = -0.903 Hz Deviation B = 0.103 - 1.0 = -0.897 Hz Deviation C = 0.092 - 1.0 = -0.908 Hz Deviation D = 0.1 - 1.0 = -0.9 Hz Deviation E = 0.102 - 1.0 = -0.898 Hz Through the above calculations, record the deviation value of each sensor, which provides a basis for subsequent timestamp calibration. After completing the sampling frequency deviation calculation, next, perform timestamp calibration on the growth environment monitoring parameters to ensure that the data of all sensors can be analyzed on a unified time scale. Suppose the sampling data of each sensor is accompanied by a timestamp in the format of "YYYY-MM-DD HH:MM:SS". Due to the sampling frequency deviation between devices, the data of some nodes may be misaligned in time. To calibrate the timestamps, first, a unified reference time needs to be set. For example, select the timestamp of the first sensor node (A) as the reference.

[0032] For each sampled data, adjust the timestamp according to its deviation value. For the first data point of Node A, its timestamp is "2025-05-12 10:00:00", while the sampling frequency deviation of Node B is -0.897 Hz, which means that the data of B lags behind the timestamp of A. Through calculation, the timestamp of Node B can be adjusted forward to "2025-05-12 10:00:00 + 0.897 seconds". For all nodes, perform timestamp calibration one by one to ensure that the monitoring parameters of each node are within a unified time scale. After completion, obtain unified time-series monitoring parameters, including the data of all sensors, which is convenient for subsequent analysis. To perform spatial repositioning, first, it is necessary to extract the spatial position coordinates of each sensor node. Assume that in a farmland, use Global Positioning System (GPS) or Geographic Information System (GIS) technology to obtain the geographical coordinates of each node. In actual operation, set the position of each sensor node, for example: Node A: (10.0, 20.0) Node B: (15.0, 25.0) Node C: (12.0, 22.0) Node D: (11.0, 21.0) Node E: (14.0, 24.0) Record these coordinates in the database for use in subsequent analysis. The spatial position data of each node can provide a basis for establishing a spatial network.

[0033] After obtaining the spatial position coordinates of each node, next, construct a spatial node coordinate network. This step aims to connect different sensor nodes through a network structure for subsequent data analysis and repositioning processing. Consider each sensor node as a node in the network, and the distance between each node can be calculated through its spatial coordinates. Use the Euclidean distance calculation method to determine the distance between Node A and other nodes: Distance between A and B: √((15.0 - 10.0)²+(25.0 - 20.0)²)=√(25 + 25)=√50≈7.07 Distance between A and C: √((12.0 - 10.0)²+(22.0 - 20.0)²)=√(4 + 4)=√8≈2.83 Distance between A and D: √((11.0 - 10.0)²+(21.0 - 20.0)²)=√(1 + 1)=√2≈1.41 Distance between A and E: √((14.0 - 10.0)²+(24.0 - 20.0)²)=√(16 + 16)=√32≈5.66 By calculating the distances between all nodes, a spatial node coordinate network is constructed to form a graph structure. The network can be represented in the form of an adjacency matrix or an edge list, which is convenient for subsequent spatial relocation analysis. Based on the spatial node coordinate network, the unified time series monitoring parameters are spatially relocated with multiple parameters to obtain the relocated standard environmental monitoring parameters. This step mainly uses the interpolation algorithm to ensure that the monitoring parameters at different locations can be reasonably relocated in space. In actual operation, Kriging interpolation, inverse distance weighted method or other spatial interpolation techniques can be used to infer the parameters of unknown points based on the monitoring parameters of known nodes. For a specific location, assuming that its distance from nodes A, B, and C is known, the monitoring parameter value of the location can be calculated by weighted average. In this way, the relocated standard environmental monitoring parameters are finally obtained, including soil moisture, conductivity, pH value, ambient temperature, humidity, and light intensity. These relocation parameters will provide a basis for subsequent growth environment monitoring and analysis, making the data more accurate and reliable.

[0034] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Obtain crop growth monitoring images throughout the entire cycle using high-definition cameras; The crop body sensor collects the crop leaf color difference, stomatal opening and closing, transpiration rate and root electrical response growth parameters to obtain the original data set of crop physiological status; Based on the original data set of crop physiological status, multi-growth stage parameters are divided to obtain original data sets of multiple growth stages; The original data sets of multiple growth stages are mined with nonlinear associations of different parameters to obtain multimodal association features; The deep evolution of growth state is carried out based on multimodal correlation characteristics, and a multimodal growth state evolution map is constructed.

[0035] In this embodiment, a high-definition camera is used to monitor the full-cycle growth of crops. The purpose of this step is to obtain visual data of crops at different growth stages, providing a basis for subsequent analysis. In the experiment, a specific crop (such as wheat or corn) is selected, and images are taken regularly during its growth cycle. It is set to take pictures once a week, and each picture includes the overall view and local details of the crop. The resolution of the high-definition camera should be at least 1080p to ensure that sufficient details are captured for subsequent analysis. The camera should be installed at a fixed position to ensure a consistent shooting angle and avoid data inconsistency caused by changes in the shooting angle. After each shooting, the images are stored in the database, and the timestamp and growth stage of each image are recorded. Assuming that 12 images are taken throughout the growth cycle, forming a complete image dataset, these images will be used for subsequent physiological state analysis and feature extraction. While obtaining the images, physiological state data of the crops are collected based on crop body sensors. The main parameters include leaf color difference, stomatal opening and closing, transpiration rate, and root electrical response, etc. These parameters can reflect the physiological state of the crops and provide necessary data support for the analysis.

[0036] In specific implementation, a color difference sensor is used to measure the color difference of the leaves, and the data recording unit is ΔE (color difference value). By measuring at different time points, the color change of the leaves can be obtained, reflecting the growth state. The opening and closing state of the stomata can be monitored in real time by a stomata sensor, recording the frequency and duration of stomatal opening and closing. The measurement of the transpiration rate is achieved by a gas flow sensor, recording the amount of water evaporated per unit time, usually in milliliters per hour (mL / h). The electrical response of the roots is measured by a resistance sensor to measure the electrical conductivity of the roots, reflecting the health state of the roots. Assume that in each growth stage, the above parameters are measured and recorded to form an original data set. After a period of monitoring, an original data set of physiological states including multiple growth stages is obtained. After completing the collection of physiological state data, the parameters of multiple growth stages are divided. The purpose of this step is to divide the entire growth cycle into multiple stages for subsequent analysis. According to the growth characteristics of the crops, the entire growth cycle is divided into several stages, such as: germination stage, seedling stage, growth stage, flowering stage, maturity stage. During the division process, it depends on the images obtained by the camera and the physiological data. The duration and characteristics of each stage can be determined by combining image analysis and sensor data. It is possible to judge whether the crop is in the flowering stage by analyzing the color change of the leaves and the opening and closing frequency of the stomata. Through data processing, the physiological parameters of each stage are extracted and an original data set of multiple stages is formed. Assume that in the data set, the color difference in the germination stage is 5.2, the color difference in the seedling stage is 7.8, and the transpiration rate in the growth stage is 15 mL / h, and the physiological parameters of each stage are recorded one by one. After completing the division of the parameters of multiple growth stages, the mining of non-linear associations of different parameters is carried out. The purpose of this step is to reveal the complex relationships between multiple physiological parameters to facilitate understanding of their impact on the growth state of the crops. For this purpose, data mining techniques such as multiple regression analysis, principal component analysis (PCA) or machine learning algorithms (such as random forests or neural networks) can be used. These methods can handle high-dimensional data and reveal the non-linear relationships between parameters. Assume that through analysis, it is found that there is a non-linear relationship between the leaf color difference and the transpiration rate. When the transpiration rate is low, the change range of the leaf color difference is large, while when the transpiration rate is high, the change range of the leaf color difference decreases. By establishing a model, the non-linear association mining of the parameters of different growth stages is carried out, and multi-modal association features are extracted. The obtained multi-modal association features can be used for further analysis to help understand the relationships between different physiological parameters and their impact on the growth state.

[0037] Deeply evolve the growth state based on multi-modal correlation features to construct a multi-modal growth state evolution map. This step aims to reveal the physiological state changes of crops at different growth stages through visual analysis. Using graph theory and visualization techniques, integrate the multi-modal correlation features to form a growth state evolution map. Each node represents a growth stage, and the connections between nodes represent the relationship of physiological state changes between different stages. The transition from the germination stage to the seedling stage can be described by the change in the stomatal opening and closing frequency and the leaf color difference. In the map, color and size can be used to represent the weights and change amplitudes of different parameters. The increase in the stomatal opening and closing frequency can be represented by the size of the node, while the change in color difference is reflected by the depth of color. In this way, the constructed evolution map can intuitively display the change process of the growth state. The evolution map will provide an important visual basis for the growth monitoring and analysis of crops, helping to optimize management strategies and improve the growth efficiency and yield of crops.

[0038] In this embodiment, refer to Figure 4 , which is a schematic diagram of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Perform adaptive region filtering and denoising on the crop full-cycle growth monitoring image stream to construct a filtered and denoised monitoring image stream; Detect crop feature points on the filtered and denoised monitoring image stream and mark multiple feature points on each image; Calculate the positions and quantities of multiple feature points on each image, and perform feature point distribution analysis to obtain the feature point characteristics of multiple images; Based on the feature point characteristics, perform crop growth cycle analysis to divide the monitoring images of different growth cycles; Perform three-dimensional morphological evolution analysis on the monitoring images of different growth cycles to obtain the three-dimensional morphological evolution law of crops.

[0039] In this embodiment, after obtaining the crop full-cycle growth monitoring image stream, first perform adaptive region filtering and denoising. This step aims to eliminate the noise in the image and improve the accuracy of subsequent feature extraction. Adaptive filtering methods (such as Wiener filtering or adaptive median filtering) are widely used in image denoising. In specific implementation, first convert the image stream into a grayscale image to simplify the processing. Perform adaptive filtering on each frame of the image. Assuming that adaptive median filtering is selected, its main steps are as follows: Window selection: Select a local window for each pixel point, and the window size is dynamically adjusted according to the noise level, usually set as a 3x3 or 5x5 area. Median calculation: Calculate the median of all pixels in the window and replace the central pixel with the median. If the median is significantly different from the central pixel, replace it with the median; otherwise, keep it unchanged. Iterative processing: Perform multiple iterations on the image to ensure that the noise is effectively removed.

[0040] In the experiment, it is assumed that the processed image stream contains 100 frames. After filtering, the noise level is significantly reduced and the image clarity is improved, which lays the foundation for subsequent feature point detection. After filtering and denoising, the crop feature points in the monitored image stream are then detected. Feature points refer to the key points that can reflect the growth state of crops, such as the leaf edges, stem nodes, etc. Feature extraction is carried out using feature point detection algorithms (such as Harris corner detection or SIFT). The specific steps are as follows: Convert each frame of the image into a grayscale image to reduce the computational complexity. Apply the Harris corner detection algorithm to calculate the response value of each pixel, and filter out the pixels with response values higher than the threshold as feature points. Mark the detected feature points on the original image with circles or crosses of different colors. In the experiment, it is assumed that the average number of feature points detected in each image is 50, and the feature points are widely distributed, which can effectively reflect the morphological characteristics of crops. After the feature point detection is completed, calculate the positions and quantities of multiple feature points in each image and conduct a distribution analysis. The purpose of this step is to understand the changes of feature points in different images. For the feature points detected in each image, record their coordinate positions (x, y) and quantities. The feature point positions of image A are [(x1, y1), (x2, y2),..., (xn, yn)]. Count the number of feature points in each image. Image A has 52 feature points and image B has 48 feature points. Analyze the distribution characteristics of the feature points and generate a feature point position distribution map. By calculating the distribution density of the feature points on the image, identify the regions where the feature points are concentrated. It is assumed that in the analysis, it is found that the feature points are mainly concentrated on the leaf edges in the early growth stage, while in the later growth stage, they extend to the stems and fruits. This distribution change provides important information for subsequent growth cycle analysis. During the division process, it depends on the change characteristics of the feature points. In the germination stage, the number of feature points is small and concentrated on the leaf edges; while in the mature stage, the number of feature points increases and the distribution becomes more uniform. Through these features, the monitored images can be effectively divided into different stages. Based on the monitored images of different growth cycles, conduct a three-dimensional morphological evolution analysis of the crops. This step aims to reveal the morphological change rules of the crops at different growth stages. Use computer vision techniques (such as structured light or multi-view stereo vision) to perform three-dimensional reconstruction of the feature points. By extracting the coordinates of the feature points from different perspectives, the construction of the three-dimensional model is realized. Analyze the changes of the three-dimensional model at different growth stages and record the key parameters (such as height, width, number of leaves, etc.). Through time series analysis techniques, extract the three-dimensional morphological evolution rules of the crops. Analyze the height growth rate, leaf expansion trend, etc. In the experiment, it is assumed that through three-dimensional analysis, it is found that the height growth rate of the crops is 1.5 cm per week during the flowering stage, and reaches 2.0 cm per week during the mature stage. These data provide a scientific basis for agricultural management and help optimize fertilization and irrigation strategies.

[0041] In this embodiment, the specific steps for adaptively filtering and denoising the image stream of the whole-cycle growth monitoring of crops and constructing a filtered and denoised monitoring image stream are as follows: Calculate the grayscale histogram of the image stream of the whole-cycle growth monitoring of crops; Perform multi-region brightness analysis on the grayscale histogram to identify over-bright or over-dark regional images; Perform histogram equalization processing on the over-bright or over-dark regional images to obtain a brightness-equalized and optimized image; Perform different-region noise analysis on the brightness-equalized and optimized image to obtain the noise characteristics of different regions; Identify the region where the crop is located based on the brightness-equalized and optimized image; perform in-depth visual recognition of the image texture details in the region where the crop is located to obtain the crop texture details; Perform adaptive region filtering and denoising on the crop texture details according to the noise characteristics of different regions to construct a filtered and denoised monitoring image.

[0042] In this embodiment, the grayscale histogram of the image stream of the whole-cycle growth monitoring of crops is calculated. The image stream consists of images at multiple time points and records the states of the crops at different growth stages. Calculating the grayscale histogram aims to analyze the distribution of grayscale values in the images, providing a basis for subsequent image processing. In specific implementation, first, each frame of the image is converted into a grayscale image. Assuming that the size of each image is 1920x1080 pixels, an image processing library (such as OpenCV) is used to convert the RGB image into a grayscale image. The grayscale value range of each pixel is from 0 to 255. Then, the number of occurrences of each grayscale value is counted, forming a histogram. Assuming that 100 frames of images are collected within a period of time, the frequencies of each grayscale value are recorded. By normalizing the grayscale histogram of each image, the sum of the histogram is made 1, facilitating the comparison of the grayscale distributions of different images. An overall histogram containing the grayscale distributions of all frame images is obtained, providing data support for subsequent brightness analysis. After obtaining the grayscale histogram, multi-region brightness analysis is performed to identify over-bright or over-dark regions in the image. This step aims to optimize the image quality and ensure the accuracy of subsequent processing. The entire image is divided into several regions. The 1920x1080 image can be divided into 16 regions (4 rows and 4 columns), and the size of each region is 480x270 pixels. For each region, its corresponding grayscale histogram is calculated, and the average brightness value of each region is analyzed. Assuming that the average grayscale value of a certain region is 240, it indicates that this region is over-bright; while the average grayscale value of another region is 20, it indicates that this region is over-dark. By setting a threshold, a grayscale value greater than 230 is regarded as over-bright, and less than 30 is regarded as over-dark. The system can automatically identify these regions. All regions marked as over-bright or over-dark are recorded, providing a basis for subsequent histogram equalization processing.

[0043] For the identified over-bright or over-dark regions, histogram equalization is performed to obtain an optimized image with balanced brightness. This process aims to improve the contrast of the image, making the details clearer. The histogram equalization algorithm in the OpenCV library is used to independently perform equalization processing on each over-bright or over-dark region. This algorithm redistributes the pixel gray values, making the gray distribution of the image more uniform. For a certain over-dark region, after equalization, a pixel with an original gray value of 20 may be adjusted to a higher gray value, thus increasing the brightness. After processing all over-bright or over-dark regions, these regions are merged with the unprocessed regions to form a new optimized image with balanced brightness. Through observation, it is assumed that after the equalization process, the contrast of the image is significantly improved and the details are clearer, providing a good basis for subsequent noise analysis and crop detection. After obtaining the optimized image with balanced brightness, noise analysis of different regions is carried out to extract the noise characteristics of different regions. Noise analysis aims to identify the interference factors in the image, thus providing a basis for subsequent denoising processing. The brightness-optimized image is evaluated using image noise assessment methods (such as root mean square error RMSE or peak signal-to-noise ratio PSNR). The image is divided into the same regions as before, and the noise level of each region is calculated. For a specific region, if its noise level is 5 dB, it indicates that the noise in this region is relatively high. By comparing the noise characteristics of different regions, the regions with larger noise are identified. These regions may affect the texture details of the crops and thus require special attention. The noise characteristics of all regions are recorded to provide data support for subsequent adaptive regional filtering denoising. Based on the brightness-optimized image, the identification of the region where the crops are located is carried out. This step aims to accurately locate the crops for subsequent texture detail analysis. Image segmentation techniques (such as threshold segmentation or semantic segmentation based on deep learning) are used to process the equalized image. By setting an appropriate threshold, the crop region is separated from the background. It is assumed that the contour of the crop region is successfully identified through the segmentation technique and marked as the target region.

[0044] For the identified crop areas, conduct in-depth visual recognition of texture details. Use texture analysis methods such as Local Binary Pattern (LBP) or Gray-Level Co-Occurrence Matrix (GLCM) to extract the texture features of the crops, which can reflect the growth status and health level of the crops. Assume that the features extracted through texture analysis include roughness and smoothness, and further analyze the relationship between these features and the growth conditions of the crops to provide data support for crop management. According to the noise characteristics of different regions, perform adaptive regional filtering denoising on the crop texture details to construct a filtered and denoised monitoring image. This step aims to improve the image quality and make subsequent analysis more accurate. For regions with high noise, adopt an adaptive filtering algorithm (such as Wiener filtering or adaptive median filtering), and adjust the filtering parameters according to the noise characteristics. For high-noise regions, set a larger window size for the filter to better remove the noise. In regions with low noise, use a smaller window to retain more details. After processing, merge all the filtered regions to obtain the final denoised monitoring image. Through observation, assume that the details of the image after denoising are clearer, and the texture features are well preserved, providing high-quality image data for subsequent crop growth monitoring and analysis.

[0045] In this embodiment, step S4 includes the following steps: According to the multi-modal growth state evolution map, conduct a correlation response analysis of the growth state parameters and three-dimensional morphology of the three-dimensional morphological evolution law of the crops to obtain the evolution response logic of growth parameters - three-dimensional morphology; Conduct a dynamic growth behavior change analysis on the evolution response logic to obtain the all-round crop growth behavior characteristics; Conduct a time-series growth evolution tracking on the all-round crop growth behavior characteristics to obtain the all-round growth evolution trajectory; According to the multi-dimensional perception model of the growth environment, conduct a digital simulation of the growth trend of the all-round growth evolution trajectory to construct a twin model of the crop growth trend.

[0046] In this embodiment, the correlation response analysis between the growth state parameters and the three-dimensional morphology is carried out according to the multi-modal growth state evolution map. This step aims to reveal the relationship between the growth parameters and the three-dimensional morphology, providing a basis for subsequent analysis. In specific implementation, first, collect the three-dimensional morphology data of the crops at different growth stages, which can be obtained through laser scanning or photogrammetry technology to obtain the three-dimensional model of the crops. Assume that in a certain growth cycle, the three-dimensional morphology data at different time points (such as sowing, emergence, heading, maturity, etc.) are collected to form a multi-modal growth state evolution map. Analyze the relationship between the growth state parameters (such as soil humidity, temperature, light intensity, and fertilizer application amount, etc.) and the three-dimensional morphology characteristics (such as height, width, leaf area, etc.). Use the correlation analysis method (such as Pearson correlation coefficient) to quantify the correlation between these parameters. Assume that the correlation between soil humidity and leaf area is found to be 0.85, indicating a strong positive correlation between the two. Through these analyses, obtain the evolution response logic between the growth parameters and the three-dimensional morphology, clarify which parameters have the greatest impact on the morphological evolution under different growth states, and provide a basis for subsequent research. After completing the correlation response analysis between the growth state parameters and the three-dimensional morphology, conduct the analysis of dynamic growth behavior changes. This step aims to deeply understand the behavior characteristics of the crops during the growth process and provide a reference for comprehensive analysis. Through time series analysis of the collected three-dimensional morphology data, extract the dynamic characteristics of the crops at different growth stages. Use the dynamic time warping (DTW) technology to compare the growth morphological changes at different time points and identify the key behavior turning points during the growth process. Assume that through analysis, it is found that the growth rate of the crops at the emergence stage is significantly higher than other stages, and this change may be related to the changes in climate conditions (such as temperature and light). Record these growth behavior characteristics and mark the important growth stages and turning points. Through this analysis, obtain the all-round growth behavior characteristics of the crops, including the growth rate, morphological changes, and their relationship with environmental factors, providing basic data for subsequent tracking of the temporal growth evolution. After obtaining the all-round growth behavior characteristics of the crops, conduct the tracking of the temporal growth evolution. This step aims to synthesize the data analyzed before to form a complete growth evolution trajectory. Based on the previously collected morphological data and growth state parameters, construct a multi-dimensional time series model, which will include the time stamps of each growth stage, the corresponding morphological characteristics, and environmental parameters. A data framework can be constructed to record the following information: Time stamp: 2025-05-01 09:00:00 Morphological characteristics: height = 10 cm, width = 5 cm, leaf area = 20 cm² Environmental parameters: soil humidity = 60%, temperature = 25 °C, light intensity = 800 µmol / m² / s; Smoothing the time series using interpolation and extrapolation methods to supplement missing data points and ensure the continuity of the growth trajectory. In this way, a complete all-round growth evolution trajectory can be obtained, facilitating subsequent digital simulation and analysis. According to the multi-dimensional perception model of the growth environment, digital simulation of the growth trend of the all-round growth evolution trajectory is carried out. This step aims to construct a digital twin model of the growth trend of crops through digital means. Input the multi-dimensional data obtained from the previous analysis into the digital model of the growth trend. The model can be trained using machine learning algorithms (such as neural networks or support vector machines), and learn the growth rules using historical data. Use the method of supervised learning to predict the output (such as the growth height and leaf area of crops) with input features (such as soil moisture, temperature, light intensity, etc.). After training, use the model to simulate the future growth trend. Assume that the simulation results show that within the next two weeks, the height of the crops will increase by 5 cm and the leaf area will increase by 10 cm. Through these simulation results, a scientific basis can be provided for crop management and decision-making. Construct a digital twin model of the growth trend of crops to ensure that the model can be updated and optimized in real time. Through continuous data collection and model iteration, the growth status of crops can be accurately monitored and predicted, providing strong support for agricultural management.

[0047] In this embodiment, step S5 includes the following steps: Conduct time evolution scale analysis based on the parameters of the crop full-cycle growth monitoring image stream to identify the actual crop growth evolution scale; Carry out multi-time scale growth simulation on the digital twin model of the crop growth trend according to the actual crop growth evolution scale to obtain long-term growth simulation data; Collect real-time monitoring parameters of the simulated growth based on the multi-dimensional crop growth environment monitoring parameters; Conduct simulation deviation analysis on the long-term growth simulation data based on the real-time monitoring parameters of the simulated growth to obtain the simulation deviation value; Conduct deviation feedback compensation on the digital twin model of the crop growth trend according to the simulation deviation value to construct a virtual-real resonance growth trend evolution model.

[0048] In this embodiment, time-evolution scale analysis is performed based on the parameters of the crop full-cycle growth monitoring image stream. The purpose of this step is to identify the growth evolution scale of real crops, so as to better understand their growth process and provide a basis for subsequent simulations. In the specific implementation process, first, collect the monitoring image streams of the crops at different growth stages. Assume that the image stream contains images taken weekly, covering the entire growth cycle (such as sowing, emergence, heading, maturity, etc.). Key parameters such as height, leaf area, and growth rate can be extracted from each image. Perform time series analysis on these parameters. Using the sliding window technique, calculate the key parameters for each growth stage to determine the growth rate and change trend. During a growth cycle, assume that the growth rates for each stage are Emergence stage: 5 cm / week Growth stage: 10 cm / week Heading stage: 3 cm / week These growth rate data will help identify the evolution scale of different stages and clarify the growth characteristics of each stage. Record the time-evolution scale of each growth stage to provide a reference basis for subsequent growth simulations. After identifying the growth evolution scale of real crops, perform multi-time scale growth simulations. This step aims to use the growth information analyzed previously to establish long-term growth simulation data. Based on the identified growth stages and rates, construct a multi-time scale growth simulation model. Assume that the model is based on dynamic system theory and considers the influence of environmental factors (such as temperature, humidity, light, etc.) on growth. Express the relationship between the growth rate and environmental parameters through equations. During the implementation process, set different time scales for simulation, for example: Short-term simulation (1 week): Use the current growth rate to directly predict the growth situation in the next week.

[0049] Medium-term simulation (1 month): Consider the periodic changes of environmental variables and use a multiple linear regression model for prediction.

[0050] Long-term simulation (3 months): Combine historical data and future climate predictions and use time series analysis (such as ARIMA model) for long-term trend prediction.

[0051] Through these simulations, long-term growth simulation data is obtained, recording information such as the growth height, leaf area, and biomass of crops at different time points. Suppose the simulation results show that the height after 3 months is 150 cm and the leaf area is 400 cm². These data will provide a basis for subsequent real-time monitoring and deviation analysis. After long-term growth simulation, real-time monitoring parameters are collected based on multi-dimensional crop growth environment monitoring parameters. This step aims to obtain real-time data corresponding to the long-term growth simulation for deviation analysis. Multiple sensor nodes are set up to collect environmental parameters such as soil humidity, temperature, light intensity, and air humidity in real time. Suppose in a farmland, the sensor collects data every 5 minutes. Through the real-time data stream, the following information is recorded: Soil humidity: 60% Temperature: 25°C Light intensity: 800 µmol / m² / s Air humidity: 70%; The real-time monitoring parameters are compared with the long-term growth simulation data to ensure data synchronization and consistency. Suppose at the same time point, the real-time monitored soil humidity matches the expected value of the simulation data. In this way, it is ensured that the growth status of crops and environmental changes can be reflected in real time. Based on the real-time monitoring parameters of the simulated growth, a simulation deviation analysis is carried out on the long-term growth simulation data. This step aims to identify the differences between the model predictions and the actual observations, and then optimize the model.

[0052] Compare the real-time monitoring parameters with the long-term simulation data to calculate the simulation deviation value. If the height predicted by the long-term simulation is 150 cm, while the actually monitored height is 145 cm, the deviation value is -5 cm. By recording the deviations at each monitoring point, a set of deviation data can be obtained. Use statistical methods (such as root mean square error RMSE or mean absolute error MAE) to quantify the overall deviation. Assume that the deviation calculation results at multiple time points are as follows: Through these analyses, evaluate the accuracy of the model and identify the main sources of deviation, such as environmental changes, inaccurate model assumptions, etc. This information provides data support for subsequent feedback compensation. Perform deviation feedback compensation on the twin model of crop growth trend according to the simulation deviation value to construct a virtual-real resonance growth trend evolution model. This step aims to optimize the prediction ability of the model through a feedback mechanism. For the identified deviations, adjust the model parameters. If certain environmental variables (such as soil moisture) have a greater impact on the model prediction, the weights of these variables in the model can be increased. Through iterative optimization algorithms (such as genetic algorithms or particle swarm optimization), continuously adjust the model parameters to reduce the deviation. Construct a virtual-real resonance growth trend evolution model, combine the real-time monitoring data with the long-term simulation data to form a dynamic feedback system. This system can update the model prediction in real time to ensure an accurate reflection of the crop growth state. Assume that after optimization, the prediction accuracy of the model is improved to over 95%, which can effectively reflect the growth trend of the crop and form a complete growth trend twin model. In this way, ensure that the growth state of the crop can be monitored and predicted in real time, providing a scientific basis for agricultural management.

[0053] In this embodiment, the specific steps of step S6 are as follows: Set different temperature and humidity and multi-frequency light perturbation sets to obtain a multi-scenario perturbation simulation matrix; Based on the multi-scenario perturbation simulation matrix, drive the growth state simulation of the virtual-real resonance growth trend evolution model to obtain the growth state simulation data under multiple perturbations; Identify the potential growth risks from the growth state simulation data to obtain the potential growth state risk points; Extract multiple parameters from the growth state simulation data and conduct a comprehensive growth state evaluation to generate a growth state simulation evaluation report; Based on the potential growth state risk points and the growth state simulation evaluation report, make an optimal intervention and regulation decision to construct an intelligent monitoring and regulation strategy for crops.

[0054] In this embodiment, different temperature, humidity, and multi - frequency light disturbance sets are set to generate a multi - scenario disturbance simulation matrix. The purpose of this step is to evaluate the impact on crop growth by simulating various environmental scenarios. In the specific implementation process, first, the ranges of temperature and humidity are determined. The temperature can be set between 15°C and 35°C, and the humidity range is 40% to 90%. At the same time, the parameters of multi - frequency light disturbance are defined. For example, the light intensity can be divided into three levels: low (200µmol / m² / s), medium (600µmol / m² / s), and high (1000µmol / m² / s). A simulation matrix is constructed. All possible combinations of temperature, humidity, and light are generated through programming languages such as Python or MATLAB. Suppose the generated matrix is as follows: Scenario 1: Temperature 20°C, Humidity 50%, Light 600µmol / m² / s Scenario 2: Temperature 30°C, Humidity 70%, Light 1000µmol / m² / s Scenario 3: Temperature 25°C, Humidity 40%, Light 200µmol / m² / s Such a simulation matrix can cover various environmental factors and provide a scenario basis for subsequent growth state simulation. After constructing the multi - scenario disturbance simulation matrix, the growth state of the virtual - real resonance growth trend evolution model is driven by simulation based on this matrix. This step aims to evaluate the growth state of crops through different environmental scenarios. The temperature, humidity, and light parameters of each scenario are input into the virtual - real resonance growth trend evolution model. The model can adopt a dynamic system model, where environmental variables affect the growth rate and growth form of crops.

[0055] Suppose the inputs of the model include: Input for Scenario 1: Temperature 20°C, Humidity 50%, Light 600µmol / m² / s; Input for Scenario 2: Temperature 30°C, Humidity 70%, Light 1000µmol / m² / s; Then, calculate the growth status under different scenarios through numerical simulation methods (such as the finite difference method or Monte Carlo simulation). The model predicts that under Scenario 1, the height of the crop increases by 3 cm, while under Scenario 2, the height increases by 5 cm. In this way, generate simulation data of the growth status under multiple perturbations, and record parameters such as the height, leaf area, and biomass of the crop corresponding to each scenario. These data will provide a basis for subsequent risk identification and status assessment. After obtaining the simulation data of the growth status, conduct potential growth risk identification. This step aims to identify potential risk points by analyzing the growth status data in order to take corresponding intervention measures. Set the criteria for risk identification. Assume that a growth rate lower than the expected value (such as 1 cm / week) or a decrease in biomass (such as less than 100 g / m²) is defined as a potential risk. Through statistical analysis of the simulation data, check whether the growth status under each scenario meets these criteria. Assume that in Scenario 3, the predicted growth rate by the model is 0.8 cm / week, and the biomass is 90 g / m², indicating that there is a potential growth risk in this scenario. Record all the identified risk points and analyze their possible causes (such as high temperature, low humidity, etc.). Through these analysis results, it can provide data support for subsequent status assessment and lay a foundation for intervention decisions. After potential growth risk identification, extract multiple parameters based on the simulation data of the growth status, and conduct a comprehensive growth status assessment to generate a simulation assessment report of the growth status.

[0056] Extract key parameters from the simulation data, such as growth height, leaf area, photosynthesis efficiency, and biomass. Use the weighted average method to calculate the comprehensive evaluation score according to the importance of different parameters. Assuming that the growth height accounts for 40%, the leaf area accounts for 30%, and the biomass accounts for 30%, a comprehensive score can be calculated. Use statistical analysis methods (such as principal component analysis PCA) to analyze the extracted parameters and identify the main factors affecting the growth state. Suppose the analysis results show that temperature and light have the greatest impact on growth. Based on the evaluation results, generate a simulation evaluation report on the growth state, which includes the growth state scores of each scenario, potential risk points, and their cause analysis. This report will provide a scientific basis for subsequent intervention and regulation decisions. According to the potential growth state risk points and the simulation evaluation report on the growth state, make the optimal intervention and regulation decisions and construct an intelligent monitoring and regulation strategy for crops. Analyze the risk points identified in the evaluation report and determine the intervention measures to be taken. If the risk in a certain scenario is caused by insufficient light, then increasing the light intensity or adjusting the planting density can be considered. Combine the real-time monitoring data and use the decision support system (DSS) to optimize the intervention strategy. By simulating the effects of different intervention schemes (such as increasing water or fertilization), select the optimal scheme. Suppose after simulation, it is decided to reduce the temperature and increase the fertilization amount in the high-temperature scenario. Construct an intelligent monitoring and regulation strategy for crops to ensure that the management measures can be flexibly adjusted in real-time monitoring. By implementing these strategies, the growth risk can be effectively reduced and the growth efficiency of crops can be improved.

[0057] In this embodiment, a crop growth monitoring system based on multi-sensor fusion is provided, which is used to execute the crop growth monitoring method based on multi-sensor fusion as described above, and includes: An environment perception module, which is used to collect multi-dimensional crop growth environment monitoring parameters, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional growth environment perception model; A growth state evolution module, which is used to obtain the full-cycle growth monitoring images of crops and the original data set of crop physiological states; perform deep evolution of the growth state according to the original data set of crop physiological states, and construct a multi-modal growth state evolution map; A morphological evolution module, which is used to perform adaptive region filtering and denoising on the full-cycle growth monitoring image stream of crops, and perform three-dimensional morphological evolution analysis of crops to obtain the three-dimensional morphological evolution law of crops; A situation simulation module, which is used to perform dynamic growth behavior change analysis on the three-dimensional morphological evolution law of crops according to the multi-modal growth state evolution map, and perform digital simulation of the growth situation according to the multi-dimensional growth environment perception model, and construct a twin model of crop growth situation; A deviation compensation module, which is used to perform multi-time scale growth simulation on the twin model of crop growth situation, and perform deviation feedback compensation to construct a virtual-real resonance growth situation evolution model; An intervention and regulation module is used to perform growth state simulation driving on the virtual-real resonance growth trend evolution model, make optimal intervention and regulation decisions, and construct an intelligent monitoring and regulation strategy for crops.

[0058] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application document are intended to be encompassed within the present invention.

[0059] As described above, these are only specific implementation manners of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A crop growth monitoring method based on multi-sensor fusion, characterized in that, It includes the following steps: Step S1: Collect multi-dimensional monitoring parameters of the crop growth environment, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional perception model of the growth environment; Step S2: Obtain the full-cycle growth monitoring images of the crops and the original dataset of the crop physiological state; perform deep evolution of the growth state based on the original dataset of the crop physiological state, and construct a multi-modal growth state evolution map; Step S3: Perform adaptive region filtering and denoising on the full-cycle growth monitoring image stream of the crops, and perform three-dimensional morphological evolution analysis of the crops to obtain the three-dimensional morphological evolution law of the crops; Step S4: Perform dynamic growth behavior change analysis on the three-dimensional morphological evolution law of the crops according to the multi-modal growth state evolution map, and perform digital simulation of the growth trend according to the multi-dimensional perception model of the growth environment, and construct a twin model of the crop growth trend; Step S5: Perform multi-time scale growth simulation on the twin model of the crop growth trend, and perform deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; Step S6: Perform growth state simulation drive on the virtual-real resonance growth trend evolution model, and perform optimal intervention and regulation decision-making to construct an intelligent monitoring and regulation strategy for crops.

2. The crop growth monitoring method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of Step S1 are: Collect multi-dimensional monitoring parameters of the crop growth environment based on multi-sensor nodes; Perform spatio-temporal standard optimization on the crop growth environment monitoring parameters to construct repositioned standard environment monitoring parameters, where the repositioned standard environment monitoring parameters include multi-dimensional soil monitoring parameters, atmospheric environment temperature and humidity parameters, and environmental real-time light intensity; Calculate the temporal fluctuations of the moisture content, conductivity, and pH value of the multi-dimensional soil monitoring parameters, and extract the soil temporal fluctuation characteristics; Mine the spatial distribution changes of the temperature and humidity of the atmospheric environment to construct a spatial distribution change map of the environmental temperature and humidity; Perform light intensity change trend analysis on the environmental real-time light intensity to obtain the environmental light intensity change trend; Perform environmental multi-dimensional perception fitting based on the soil temporal fluctuation characteristics, environmental light intensity change trend, and spatial distribution change map of the environmental temperature and humidity, and construct a multi-dimensional perception model of the growth environment.

3. The crop growth monitoring method based on multi-sensor fusion according to claim 2, wherein, The specific steps of performing spatio-temporal standard optimization on the crop growth environment monitoring parameters to construct repositioned standard environment monitoring parameters are: Calculate the sampling frequency deviation between devices of the multi-sensor nodes, and extract the multi-sensor sampling frequency deviation; Calibrate the timestamps of the growth environment monitoring parameters according to the multi-sensor sampling frequency deviation to obtain unified time-series monitoring parameters; Calculate the spatial positions of the multi-sensor nodes, and extract the position coordinates of each node; Construct a spatial node coordinate network according to the position coordinates of each node; Perform multi-parameter spatial repositioning on the unified time-series monitoring parameters based on the spatial node coordinate network to obtain repositioned standard environment monitoring parameters.

4. The crop growth monitoring method based on multi-sensor fusion according to claim 1, wherein The specific steps of Step S2 are: Obtain the full-cycle growth monitoring images of the crops according to the high-definition camera; Collect the growth parameters of the crop leaf color difference, stomatal opening and closing, transpiration rate, and root electrical response based on the crop body sensors to obtain the original dataset of the crop physiological state; Based on the original dataset of crop physiological states, perform parameter division for multiple growth stages to obtain the original datasets for multiple growth stages; Perform non-linear association mining of different parameters on the original datasets for multiple growth stages to obtain multi-modal association features; Based on the multi-modal association features, perform in-depth evolution of the growth state and construct a multi-modal growth state evolution map.

5. The crop growth monitoring method based on multi-sensor fusion according to claim 1, characterized in that The specific steps of step S3 are as follows: Perform adaptive region filtering and denoising on the crop full-cycle growth monitoring image stream to construct a filtered and denoised monitoring image stream; Perform crop feature point detection on the filtered and denoised monitoring image stream, and mark multiple feature points on each image; Calculate the positions and quantities of multiple feature points on each image, and perform feature point distribution analysis to obtain the feature point characteristics of multiple images; Based on the feature point characteristics, perform crop growth cycle analysis and divide the monitoring images of different growth cycles; Perform crop three-dimensional morphology evolution analysis on the monitoring images of different growth cycles to obtain the crop three-dimensional morphology evolution law.

6. The crop growth monitoring method based on multi-sensor fusion according to claim 5, wherein, The specific steps of performing adaptive region filtering and denoising on the crop full-cycle growth monitoring image stream to construct a filtered and denoised monitoring image stream are as follows: Calculate the gray histogram of the crop full-cycle growth monitoring image stream; Perform multi-region brightness analysis on the gray histogram to identify over-bright or over-dark regional images; Perform histogram equalization processing on the over-bright or over-dark regional images to obtain a brightness-balanced optimized image; Perform different-region noise analysis on the brightness-balanced optimized image to obtain the noise characteristics of different regions; Identify the region where the crop is located according to the brightness-balanced optimized image; Perform in-depth visual recognition of the image texture details in the region where the crop is located to obtain the crop texture details; Perform adaptive region filtering and denoising on the crop texture details according to the noise characteristics of different regions to construct a filtered and denoised monitoring image.

7. The crop growth monitoring method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of step S4 are as follows: According to the multi-modal growth state evolution map, perform correlation response analysis of the growth state parameters and three-dimensional morphology of the crop three-dimensional morphology evolution law to obtain the evolution response logic of growth parameters - three-dimensional morphology; Perform dynamic growth behavior change analysis on the evolution response logic to obtain all-round crop growth behavior characteristics; Perform time-series growth evolution tracking on the all-round crop growth behavior characteristics to obtain an all-round growth evolution trajectory; According to the multi-dimensional growth environment perception model, perform digital simulation of the growth trend on the all-round growth evolution trajectory to construct a crop growth trend twin model.

8. The crop growth monitoring method based on multi-sensor fusion according to claim 1, characterized in that The specific steps of step S5 are as follows: Perform time evolution scale analysis according to the crop full-cycle growth monitoring image stream parameters to identify the actual crop growth evolution scale; Perform multi-time scale growth simulation on the crop growth trend twin model according to the actual crop growth evolution scale to obtain long-term growth simulation data; Collect real-time monitoring parameters of simulated growth based on multi-dimensional crop growth environment monitoring parameters; Perform simulation deviation analysis on the long-term growth simulation data based on the real-time monitoring parameters of the simulated growth to obtain a simulation deviation value; Perform deviation feedback compensation on the crop growth trend twin model according to the simulation deviation value to construct a virtual-real resonance growth trend evolution model.

9. The crop growth monitoring method based on multi-sensor fusion according to claim 1, characterized in that The specific steps of step S6 are as follows: Set different temperature, humidity and multi-frequency light disturbance sets to obtain a multi-scenario disturbance simulation matrix; Based on the multi-scenario disturbance simulation matrix, perform growth state simulation driving on the virtual-real resonance growth trend evolution model to obtain growth state simulation data under multiple disturbances; Identify potential growth risks from the growth state simulation data to obtain potential growth state risk points; Extract multiple parameters based on the growth state simulation data, and conduct a comprehensive growth state evaluation to generate a growth state simulation evaluation report; Based on the potential growth state risk points and the growth state simulation evaluation report, make an optimal intervention and regulation decision to construct an intelligent monitoring and regulation strategy for crops.

10. A crop growth monitoring system based on multi-sensor fusion, characterized in that, Used to execute the crop growth monitoring method based on multi-sensor fusion as described in claim 1, including: An environmental perception module, which is used to collect multi-dimensional crop growth environment monitoring parameters, perform environmental multi-dimensional perception fitting, and construct a multi-dimensional growth environment perception model; A growth state evolution module, which is used to obtain the full-cycle growth monitoring images of crops and the original dataset of crop physiological states; perform deep evolution of the growth state based on the original dataset of crop physiological states, and construct a multi-modal growth state evolution map; A morphological evolution module, which is used to perform adaptive region filtering and denoising on the full-cycle growth monitoring image stream of crops, and conduct three-dimensional morphological evolution analysis of crops to obtain the three-dimensional morphological evolution law of crops; A trend simulation module, which is used to analyze the dynamic growth behavior changes of the three-dimensional morphological evolution law of crops according to the multi-modal growth state evolution map, and perform digital simulation of the growth trend according to the multi-dimensional growth environment perception model to construct a crop growth trend twin model; A deviation compensation module, which is used to perform multi-time scale growth simulation on the crop growth trend twin model, and perform deviation feedback compensation to construct a virtual-real resonance growth trend evolution model; An intervention and regulation module, which is used to perform growth state simulation driving on the virtual-real resonance growth trend evolution model, and make an optimal intervention and regulation decision to construct an intelligent monitoring and regulation strategy for crops.

Images