Analysis and decision method for crop growth monitoring data

Abstract

The invention relates to the technical field of data decision, in particular to an analysis decision method for crop growth monitoring data. The method comprises the following steps: collecting crop phenotypic characteristic data through a multi-source sensing array; dynamically dividing crop growth response stages according to phenotypic characteristics; constructing a growth situation prediction model based on the crop growth response stage; inputting the phenotypic characteristic data into a growth situation prediction model, simulating a crop growth process through a digital twin engine, and generating growth situation evolution data; extracting physiological anomaly characteristics of the growth situation evolution data, and generating stress response signals; obtaining a canopy three-dimensional phenotype map based on the stress response signal; positioning vascular bundle nodes of the canopy three-dimensional phenotype map through morphological topology analysis; according to the invention, through multi-source data acquisition, dynamic growth stage division, accurate model prediction, three-dimensional phenotype monitoring and self-correction decision optimization, the accuracy of analysis decision for crop growth monitoring data is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data decision-making, and in particular, to an analysis and decision-making method for crop growth monitoring data. Background Art

[0002] Crop growth monitoring relies on manual observation and traditional agricultural experience. This method is not only inefficient but also limited by subjective factors, making it impossible to achieve precise agricultural decision-making. With the development of remote sensing technology, crop growth monitoring has started to use satellite images and drone images, combined with ground meteorological data and soil information, to achieve remote monitoring. The monitoring methods at this stage can infer the growth of crops by analyzing features such as reflectivity and vegetation indices (such as NDVI) in image data. After entering the 21st century, with the development of technologies such as big data, cloud computing, and the Internet of Things, the data sources for crop growth monitoring have become more diverse and accurate. Sensor technology enables real-time data such as soil humidity, temperature, and light to be directly transmitted to the decision-making system, forming a more comprehensive monitoring network. In addition, the application of machine learning and deep learning technologies enables more precise crop growth prediction and pest and disease early warning through the combination of historical data and real-time monitoring data. However, currently, traditional growth prediction models often fail to fully consider the details of crop response to external stresses, resulting in insufficient prediction accuracy. At the same time, it is difficult for traditional methods to detect the blockage phenomenon in the photosynthetic assimilation process in real time and accurately. Summary of the Invention

[0003] Based on this, it is necessary to provide an analysis and decision-making method for crop growth monitoring data to solve at least one of the above technical problems.

[0004] To achieve the above object, an analysis and decision-making method for crop growth monitoring data, the method includes the following steps: Step S1: Collect crop phenotypic characteristic data through a multi-source sensing array; dynamically divide the crop growth response stage according to the phenotypic characteristics; construct a growth trend prediction model based on the crop growth response stage; Step S2: Input the phenotypic characteristic data into the growth trend prediction model, and simulate the crop growth process through a digital twin engine to generate growth trend evolution data; extract the physiological abnormal characteristics of the growth trend evolution data to generate stress response signals; Step S3: Obtain a three-dimensional canopy phenotypic map based on the stress response signal; locate the vascular bundle nodes of the three-dimensional canopy phenotypic map through morphological topology analysis; obtain a phloem microscopic image based on the node coordinates and perform chloroplast fluid state detection. If it is detected that the fluid state entropy value exceeds the preset threshold interval, it is determined as a photosynthetic assimilation blockage phenomenon; Step S4: Analyze the regulatory defects of the photosynthetic assimilation blockage phenomenon, and perform parameter self-correction on the growth trend prediction model through the regulatory defects to generate an optimized growth decision-making model.

[0005] The present invention collects phenotypic characteristic data of crops through a multi-source sensor array, and then dynamically divides the crop growth response stage, which provides basic data for the subsequent growth trend prediction model. This step can achieve precise crop growth monitoring and provide data support for different growth stages, thereby improving the accuracy of prediction. Input the collected phenotypic characteristic data into the growth trend prediction model, use the digital twin engine to simulate the growth process of the crop, and generate growth trend evolution data. Further extract the physiological abnormal characteristics in these data to identify potential stress response signals. Through digital twin technology, it is possible to simulate each stage of crop growth in real time, thereby early identifying physiological abnormalities and timely detecting the stresses suffered by the crops (such as water, nutrients or environmental stress), providing support for precision agriculture. Based on the stress response signal, obtain the three-dimensional phenotypic map of the crop canopy, and locate the vascular bundle nodes through morphological topology analysis. These node information will be used to obtain phloem microscopic images and perform chloroplast fluid state detection. If the fluid state entropy value exceeds the preset threshold range, there is a photosynthetic assimilation blockage phenomenon. The three-dimensional phenotypic map and fluid state detection technology can deeply analyze the photosynthesis process of the crop, timely discover changes in photosynthetic assimilation efficiency, and provide a scientific basis for agronomic regulation. Analyze the regulatory defects of the photosynthetic assimilation blockage phenomenon, and perform self-correction on the growth trend prediction model based on these defects, thereby generating an optimized growth decision-making model. By self-regulating the model through the feedback mechanism, it is possible to more precisely adjust the crop growth decision-making and improve the scientificity and effectiveness of crop management. Therefore, the present invention improves the accuracy of the analysis and decision-making for crop growth monitoring data through multi-source data acquisition, dynamic growth stage division, precise prediction model, three-dimensional phenotypic monitoring and self-correction decision optimization.

[0006] Preferably, step S1 includes the following steps: Step S11: Collect crop phenotypic characteristic data through a multi-source sensing array, where the crop phenotypic characteristic data includes canopy spectral reflectance data and rhizosphere microenvironment data; Step S12: Calculate the NDVI index for the canopy spectral reflectance data to obtain the crop growth condition index; Step S13: Extract the moisture and temperature of the rhizosphere microenvironment data, and combine the crop growth condition index to analyze the root prosperity degree of the crop to generate crop root prosperity degree data; Step S14: Divide the crop growth response stage based on the crop root prosperity degree data to obtain the crop growth response stage; Step S15: Construct a growth trend prediction model according to the crop growth response stage.

[0007] Through the integration of multi-source data, the present invention can capture different dimensions of crop growth more comprehensively, providing rich data support for subsequent analysis. Canopy spectral reflectance data helps monitor the photosynthesis status, while rhizosphere microenvironment data can reflect the impact of soil conditions on crop root growth. NDVI is an important indicator for monitoring crop health and growth status. Through this indicator, the photosynthesis intensity, health status, and growth rate of crops can be quickly evaluated, providing intuitive data for the analysis of crop growth status. The health and development of roots are crucial for crop growth. By combining rhizosphere moisture, temperature data, and crop growth status, the prosperity of roots can be accurately evaluated, and then the water and nutrient absorption capabilities of crops, as well as their adaptability to environmental changes, can be predicted. By dividing different growth stages of crops, more targeted management can be carried out according to the characteristics and requirements of crops in each stage. The division of growth stages can help farmers perform operations such as irrigation, fertilization, and pest control at the appropriate time, thereby improving the yield and quality of crops. This prediction model can accurately predict the future growth trend of crops based on real-time collected growth data. The model can provide decision-making support for agricultural management and optimize strategies such as fertilization, irrigation, and pest control.

[0008] Preferably, step S15 includes the following steps: Step S151: Collect historical crop stage-matched growth data based on the crop growth response stage; Step S152: Divide the historical crop stage-matched growth data into data sets to generate a model training set and a model test set; Step S153: Train the model training set through the support vector machine algorithm to generate a preliminary growth trend prediction model; Step S154: Use the model test set to optimize and iterate the preliminary growth trend prediction model to generate a growth trend prediction model.

[0009] By introducing historical data, the present invention can provide a rich sample library for the prediction model. These data reflect the actual performance of crops at different growth stages, enhancing the learning basis of the model. Historical data helps the model capture the patterns and trends in the crop growth process, thereby improving the prediction accuracy of the model. The division of the dataset is a crucial step in machine learning, which ensures the generalization ability of the model. While training on the training set, the test set is used to evaluate the performance of the model, thus avoiding the overfitting problem. This can ensure that the trained model has good prediction performance in practical applications. The support vector machine algorithm is a very effective supervised learning method, especially suitable for classification and regression problems. Through the SVM algorithm, the model can find the optimal hyperplane to separate the data of different growth stages, thus accurately learning the crop growth patterns and rules. The trained pre-model can capture the characteristics of crop growth at different stages, laying a foundation for subsequent predictions. By optimizing and iterating on the test set, the accuracy and robustness of the model can be further improved. Optimization and iteration can help the model adjust parameters and reduce prediction errors, thereby generating a more accurate and adaptable growth trend prediction model. The optimized model can better handle new data and enhance its feasibility in practical applications.

[0010] Preferably, step S2 includes the following steps: Step S21: Input the phenotypic feature data into the growth trend prediction model for crop growth trend prediction, and generate crop growth trend prediction data; Step S22: Simulate the crop growth process for the growth trend prediction data through the digital twin engine, and generate growth trend evolution data; Step S23: Extract the physiological abnormality features of the growth trend evolution data, and perform time series analysis on the physiological abnormality features to generate a physiological abnormality time window; Step S24: Quantify the stress response index for the growth trend evolution data according to the physiological abnormality time window, and generate a stress response index; issue a signal warning for the physiological abnormality features through the stress response index, and generate a stress response signal.

[0011] By inputting multi-source phenotypic feature data into a prediction model, the present invention can achieve dynamic prediction of crop growth status and predict the future growth trend of crops in advance. Through this process, real-time crop growth prediction can be provided for agricultural managers, providing data support for subsequent management decisions such as fertilization and irrigation. Digital twin technology can establish a virtual model for the crop growth process, simulating and reflecting the growth evolution of crops in real time. This real-time simulation can help agricultural managers more intuitively understand the dynamic process of crop growth and the potential problems faced, and then formulate optimized planting plans. By extracting physiological anomaly features, abnormal changes in crop growth (such as excessive water, high temperature, nutrient deficiency, etc.) can be identified. These changes are early signals of crop stress responses. Time series analysis helps track the duration and frequency of these abnormal changes, providing a basis for subsequent response measures. The stress response index quantifies the intensity of the crop's response to environmental stress, helping farmers and agricultural experts timely understand the response degree of crops when facing stress (such as drought, pests and diseases, etc.). Through signal warning, response measures (such as adjusting irrigation, fertilization, pest and disease control, etc.) can be taken in advance to reduce crop losses and optimize the agricultural management process.

[0012] Preferably, the steps for extracting the physiological anomaly features of the growth trend evolution data in step S23 include the following steps: When any of the following situations occurs, the growth trend evolution data is determined to be abnormal in growth trend evolution, and growth trend evolution abnormal data is obtained: the growth rate deviates from the optimal range by more than ±10% cm / day; the root growth length is lower than 10 cm or higher than 50 cm; the environmental temperature deviates from the normal range of 18 - 30 °C; the light intensity is lower than 500 lux or higher than 2000 lux; When the following situations occur simultaneously, the growth trend evolution data is determined to be physiologically dysfunctional and physiologically dysfunctional data is obtained: the photosynthesis rate is lower than 20 μmol / m² / s for 3 consecutive days; the leaf water content is lower than 40%; the pH value deviates from the range of 5.5 - 7.5; the soil nutrient concentration is lower than 50% mg / kg of the normal level; When the following situations occur simultaneously, the growth trend evolution data is determined to be a water absorption failure and water absorption failure data is obtained: the root water absorption rate continuously decreases by more than 20% ml / hour; the plant transpiration rate is lower than 30% g / hour of the normal level; the soil humidity deviates from the range of 40 - 60%; the plant leaves turn yellow and cannot recover in time; Integrate the growth trend evolution abnormal data, physiologically dysfunctional data, and water absorption failure data to obtain the physiological anomaly features of the growth trend evolution data.

[0013] By setting specific numerical thresholds (such as light intensity, temperature, photosynthesis rate, etc.), the present invention can effectively identify different types of physiological abnormalities, including abnormal growth trend evolution, physiological dysfunction, and water absorption failure. Timely feedback of these abnormal data can provide early warning for plant growth management, avoid further development of problems, and reduce losses. This method integrates multiple indicators and can comprehensively and meticulously understand the growth state of plants through integrated analysis of data, helping to improve agricultural planting efficiency. It can support the development of an intelligent agricultural management system, realize precision agriculture based on data, and enhance the ability to regulate the plant growth environment. By identifying and locating different types of physiological abnormalities, it can provide an effective basis for agricultural decision-making, contribute to adjusting the planting plan or optimizing environmental conditions, and ensure the healthy growth of plants.

[0014] Preferably, the obtaining of the three-dimensional canopy phenotypic map based on the stress response signal in step S3 includes: Locating the regional position coordinates of the crop canopy based on the stress response signal; Scanning the regional position coordinates of the crop canopy with a spectral lidar array to obtain crop canopy point cloud data and reflectivity; Extracting the geometric topology of the crop canopy point cloud data and setting the voxel size to 2 cm³ to construct a voxel space, obtaining crop canopy voxel modeling data; Screening the abnormal reflection regions of the crop canopy voxel modeling data according to the reflectivity, and performing chlorophyll content inversion on the abnormal reflection regions to generate a three-dimensional canopy phenotypic map.

[0015] Through the precise positioning of stress response signals, the present invention can ensure that the regional position coordinates of the canopy are accurately determined, laying a foundation for subsequent point cloud data acquisition and three-dimensional modeling. After accurately positioning the canopy area, the subsequent point cloud data acquisition and modeling work will be more efficient and accurate, avoiding data errors caused by position deviations. The spectral lidar array provides a detailed scan of the crop canopy, enabling rapid collection of spatial point cloud data on the crop surface and providing precise geometric data for three-dimensional modeling. Reflectance data can reflect the optical properties of the crop canopy, providing a reliable basis for subsequent screening of abnormal areas and inversion of chlorophyll content. The spectral lidar scanning technology can avoid the errors of traditional shooting methods and ensure the accuracy and consistency of data. By extracting the geometric structure of the point cloud data, the spatial morphology of the crop canopy can be accurately constructed, and the detailed features of the canopy can be identified. Using a voxel size of 2 cm³ helps to ensure the accuracy and details of the modeling, clearly showing the three-dimensional structure of the crop canopy and providing support for subsequent map generation and data analysis. Through the construction of the voxel space, large-scale three-dimensional data can be effectively processed and stored, reducing data redundancy and improving computational efficiency. Screening abnormal reflection areas can help identify areas with growth problems in the crop canopy, such as areas affected by stress or diseases. The identification of these areas helps to carry out targeted intervention and management. By screening abnormal reflection areas, problems can be located more efficiently, saving manual analysis time and improving the accuracy of diagnosis. Inverting the chlorophyll content can reveal the crop health status, photosynthetic capacity, and the intensity of stress response, especially for evaluating plant physiological changes. By generating a three-dimensional canopy phenotypic map, not only the external morphology of the crop can be presented, but also the internal physiological characteristics can be revealed, providing detailed data support for crop health monitoring.

[0016] Preferably, the positioning of the vascular bundle nodes of the three-dimensional canopy phenotypic map through morphological topology analysis in step S3 includes: Extracting the local geometric features of the three-dimensional canopy phenotypic map to obtain curvature, normal vectors, and neighborhood relationships; Performing density clustering on the three-dimensional canopy phenotypic map based on the curvature, normal vectors, and neighborhood relationships to generate feature clusters; Using the feature clusters to screen candidate vascular bundle regions in the three-dimensional canopy phenotypic map and performing connectivity analysis on the candidate vascular bundle regions to obtain regional connectivity; Performing branch structure analysis on the candidate vascular bundle regions to obtain regional branch data; Performing vascular bundle morphological topology screening on the candidate vascular bundle regions based on the regional connectivity and regional branch data to obtain the vascular bundle nodes of the three-dimensional canopy phenotypic map.

[0017] Through curvature and normal analysis, the present invention can accurately describe the local geometric morphology of the canopy, providing key features for subsequent topological analysis. The extraction of neighborhood relationships helps to better understand the mutual relationships and spatial layout among points within the canopy, thereby enhancing the detail representation ability of the canopy phenotypic map. Density clustering can group regions with similar geometric features into one set, making the data structure more organized and hierarchical, facilitating subsequent processing. Through clustering, the vascular bundle regions in the canopy can be efficiently identified, reducing redundant data and improving the analysis efficiency. Connectivity analysis helps to determine whether the structure within the candidate region is complete, thereby ensuring the accuracy of the vascular bundle region. By analyzing regional connectivity, the structural breakpoints can be effectively identified, providing support for subsequent branch structure analysis and vascular bundle morphology screening. Branch structure analysis can reveal the growth pattern and vascular bundle distribution of the crop canopy, thereby providing key data for aspects such as the growth process and stress response of the crop. By identifying the branch structure, the distribution of vascular bundle nodes can be more accurately located, further improving the extraction accuracy of vascular bundle nodes. Through morphological topology screening, the positions of vascular bundle nodes can be accurately identified and located, thereby providing detailed data for research in aspects such as crop growth and nutrient transport. The accurate identification of vascular bundle nodes helps to establish a more biologically significant crop growth model, enabling researchers to more deeply understand the growth mechanism of the crop, especially for the analysis of processes such as water and nutrient absorption of the crop.

[0018] Preferably, the obtaining of the phloem microscopic image based on the node coordinates and the chloroplast fluid state detection in step S3 includes: Calculating the node coordinates of the vascular bundle nodes of the three-dimensional phenotypic map of the canopy, and performing microscopic photography on the vascular bundle region according to the node coordinates to obtain the phloem microscopic image; Performing chloroplast detection processing on the phloem microscopic image to generate chloroplast fluid state region data; Extracting the fluid state features of the chloroplast fluid state region data, and calculating the information entropy of the chloroplast fluid state region data to obtain the fluid state entropy value; Comparing the fluid state entropy value with a preset threshold interval. When the fluid state entropy value is greater than the preset threshold interval, the fluid state entropy value is marked as the phenomenon of photosynthetic assimilation blockage.

[0019] Through precise calculation of the coordinates of vascular bundle nodes, the present invention can effectively locate the area to be photographed, ensuring that the obtained images are representative and cover key growth nodes. Guiding microscopic photography through node coordinates helps reduce shooting errors and improve the accuracy and quality of the obtained phloem microscopic images. The detection and processing of chloroplasts can automatically identify and locate chloroplasts in the images, reducing manual intervention and enhancing processing efficiency. Accurately extracting the fluid state area of chloroplasts helps further analyze the physiological functions of crops and the efficiency of their photosynthesis. By extracting fluid state characteristics, the fluid state of chloroplasts can be accurately characterized, and thus the photosynthesis characteristics of crops can be understood. The calculation of information entropy provides a quantitative means to characterize the complexity and changes of chloroplast fluid state, providing quantitative support for subsequent analysis. By comparing the fluid state entropy values, the phenomenon of photosynthetic assimilation blockage can be effectively identified, and early warnings can be given for the growth problems of crops. Threshold comparison provides a quantitative standard for the identification of photosynthetic assimilation blockage, avoiding errors in subjective judgment and improving the reliability of diagnostic results.

[0020] Preferably, step S4 includes the following steps: Step S41: Collect carbon flux deficit data of the blockage event based on the photosynthetic assimilation blockage phenomenon; Step S42: Calculate the elastic coefficient of the rate-limiting enzyme for the carbon flux deficit data through metabolic control analysis; Step S43: Determine the expression levels of Rubisco activase and SPS proteins and construct an enzyme kinetics correction function; set the starch / sucrose output ratio constraint condition for the elastic coefficient of the rate-limiting enzyme through the elastic coefficient of the rate-limiting enzyme; Step S44: Use the starch / sucrose output ratio constraint condition to perform parameter self-correction on the growth trend prediction model to generate an optimized growth decision model.

[0021] By quantifying the carbon flux deficit data, the present invention can accurately monitor carbon loss during the photosynthetic assimilation process and help study the loss of photosynthesis efficiency. The carbon flux deficit data provides important input data for subsequent metabolic control analysis, further guiding the optimization of plant physiological characteristics. By calculating the elastic coefficient of the rate-limiting enzyme, the blocked process of carbon flux can be deeply understood, revealing the key role of specific enzymes in photosynthetic assimilation blockage. The elastic coefficient of the rate-limiting enzyme quantifies the limiting effect of the enzyme in photosynthesis, helping to specifically improve the metabolic process of crops. By constructing the enzyme kinetics correction function, the metabolic regulation of crops can be effectively optimized and the photosynthesis efficiency can be enhanced. By setting the starch and sucrose output ratio constraint condition, it helps to regulate the carbon storage and transportation processes of crops and improve the growth efficiency and stress resistance of crops. By self-correcting the growth trend prediction model, the prediction accuracy can be significantly improved, and the strategies for crop growth can be optimized. The optimized growth decision model can dynamically adjust agricultural management measures according to the growth conditions of crops, achieving the goal of precision agriculture.

[0022] Preferably, step S44 includes the following steps: Step S441: Determine the starch / sucrose output ratio. When the starch / sucrose output ratio is less than 0.5, generate the first constraint condition; when the starch / sucrose output ratio is between 0.5 and 1.5, generate the second constraint condition; when the starch / sucrose output ratio is greater than 1.5, generate the third constraint condition; Step S442: Perform the first type of parameter self - calibration on the growth trend prediction model based on the first constraint condition to obtain the first constraint adjustment parameter. The first type of parameter self - calibration includes increasing the photosynthetic rate parameter in the growth trend prediction model to 0.5 μmol / m² / s, increasing the starch synthesis rate to 0.2 mol / h, reducing the sucrose transport rate to 0.05 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S443: Perform the second type of parameter self - calibration on the growth trend prediction model based on the second constraint condition to obtain the second constraint adjustment parameter. The second type of parameter self - calibration includes maintaining the starch synthesis rate in the growth trend prediction model at 0.15 mol / h, maintaining the sucrose synthesis rate at 0.1 mol / h, keeping the sucrose transport rate at 0.05 mol / h, maintaining the sucrose consumption rate at 0.05 mol / h, and keeping the starch storage rate and sucrose utilization rate within the preset optimal range without additional adjustment; Step S444: Perform the first type of parameter self - calibration on the growth trend prediction model based on the third constraint condition to obtain the third constraint adjustment parameter. The third type of parameter self - calibration includes reducing the starch synthesis rate in the growth trend prediction model to 0.2 mol / h, increasing the sucrose synthesis rate to 0.15 mol / h, increasing the sucrose transport rate to 0.1 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S445: Integrate the first constraint adjustment parameter, the second constraint adjustment parameter, and the third constraint adjustment parameter into a parameter self - calibration strategy, and use the parameter self - calibration strategy to perform decision training on the growth trend prediction model to generate an optimized growth decision model.

[0023] By judging the starch / sucrose output ratio, different constraint conditions are set for different crop growth stages and environmental conditions to ensure that the model can make reasonable adjustments in various situations. Adjusting the growth model according to different situations of the output ratio can carry out refined management for different growth requirements of crops, improve the controllability of crop growth and the prediction accuracy. When the starch / sucrose output ratio is lower than 0.5, the increase in photosynthetic rate and starch synthesis rate can promote crop growth, improve carbon metabolism, and enhance the stress tolerance of crops. By increasing the photosynthetic rate and regulating the starch and sucrose synthesis rates, it helps to enhance the carbon accumulation of crops, especially when photosynthesis is limited, to ensure the normal growth of plants. When the output ratio is between 0.5 and 1.5, ensure that each growth parameter is within a reasonable range, maintain the stability of growth, and there is no need for large adjustments, ensuring the sustainability and health of crop growth. By maintaining a balanced synthesis and consumption rate, avoid the state of overconsumption or insufficient storage of crops, and improve the overall metabolic efficiency of crops. When the starch / sucrose output ratio is greater than 1.5, by adjusting the synthesis and consumption rates of sucrose and starch, it can promote the normal growth of crops under high metabolic demands and enhance the effective utilization of carbon. By increasing the synthesis and transport rate of sucrose and adjusting the starch synthesis rate, the distribution of carbon in the plant can be made more reasonable, and the effective utilization of metabolites after photosynthesis can be improved. By integrating self-correcting parameters under different constraint conditions, the crop growth prediction model can be comprehensively optimized, and the adaptability and accuracy of the model can be improved. The optimized model can provide a more accurate basis for agricultural management decisions, support the refined management of crop growth, and enhance the stability and efficiency of agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a schematic flow chart of the steps of an analysis and decision-making method for crop growth monitoring data; Figure 2 is Figure 1 a detailed implementation step flow chart of step S2 in; Figure 3 is Figure 1 a detailed implementation step flow chart of step S4 in; The realization, functional features and advantages of the object of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that although terms such as "first" and "second" may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve the above object, please refer to Figures 1 to 3 , an analysis and decision-making method for crop growth monitoring data, the method comprising the following steps: Step S1: Collect crop phenotypic characteristic data through a multi-source sensing array; dynamically divide the crop growth response stage according to the phenotypic characteristics; construct a growth trend prediction model based on the crop growth response stage; Step S2: Input the phenotypic characteristic data into the growth trend prediction model, and simulate the crop growth process through a digital twin engine to generate growth trend evolution data; extract the physiological abnormality characteristics of the growth trend evolution data to generate stress response signals; Step S3: Obtain a three-dimensional canopy phenotypic map based on the stress response signal; locate the vascular bundle nodes of the three-dimensional canopy phenotypic map through morphological topology analysis; obtain a phloem microscopic image based on the node coordinates and perform chloroplast fluid state detection. If the detected fluid state entropy value exceeds the preset threshold range, it is determined as a photosynthetic assimilation blockage phenomenon; Step S4: Analyze the regulation defects of the photosynthetic assimilation blockage phenomenon, and perform parameter self-correction on the growth trend prediction model through the regulation defects to generate an optimized growth decision-making model.

[0029] The present invention collects phenotypic characteristic data of crops through a multi-source sensor array, and then dynamically divides the crop growth response stage, which provides basic data for the subsequent growth trend prediction model. This step can achieve precise crop growth monitoring and provide data support for different growth stages, thereby improving the accuracy of prediction. The collected phenotypic characteristic data is input into the growth trend prediction model, and the digital twin engine is used to simulate the growth process of the crops and generate growth trend evolution data. Further, the physiological abnormality characteristics in these data are extracted to identify potential stress response signals. Through digital twin technology, each stage of crop growth can be simulated in real time, so as to identify physiological abnormalities at an early stage and timely detect the stresses (such as water, nutrients or environmental pressure) suffered by the crops, providing support for precision agriculture. Based on the stress response signals, a three-dimensional phenotypic map of the crop canopy is obtained, and the vascular bundle nodes are located through morphological topology analysis. These node information will be used to obtain phloem microscopic images and conduct chloroplast fluid state detection. If the fluid state entropy value exceeds the preset threshold range, there is a phenomenon of photosynthetic assimilation blockage. The three-dimensional phenotypic map and fluid state detection technology can deeply analyze the photosynthesis process of crops, timely discover changes in photosynthetic assimilation efficiency, and provide a scientific basis for agronomic regulation. Analyze the regulatory defects of the photosynthetic assimilation blockage phenomenon, and self-correct the growth trend prediction model based on these defects, thereby generating an optimized growth decision model. By self-regulating the model through a feedback mechanism, the crop growth decision can be adjusted more precisely, improving the scientificity and effectiveness of crop management. Therefore, the present invention improves the accuracy of the analysis and decision-making for crop growth monitoring data through multi-source data acquisition, dynamic growth stage division, precise prediction model, three-dimensional phenotypic monitoring and self-correcting decision optimization.

[0030] In an embodiment of the present invention, with reference to Figure 1 as shown, it is a schematic flow chart of the steps of an analysis and decision-making method for crop growth monitoring data according to the present invention. In this example, the analysis and decision-making method for crop growth monitoring data includes the following steps: Step S1: Collect crop phenotypic characteristic data through a multi-source sensing array; dynamically divide the crop growth response stage according to the phenotypic characteristics; construct a growth trend prediction model based on the crop growth response stage; In the embodiments of the present invention, by using a variety of sensor arrays, including spectral sensors, infrared sensors, hyperspectral sensors, visible light cameras, lidar (LiDAR), etc., crop growth information in different dimensions is captured. Different crops are regularly monitored through drones, ground sensors or satellite platforms to ensure the spatio-temporal continuity of data. Data collection should cover different growth stages of crops, such as sowing stage, seedling stage, tillering stage, heading stage, filling stage and maturity stage. Methods such as time series analysis, clustering analysis, machine learning models, etc. are used to automatically determine the transition points of growth stages according to the growth dynamics and changes in phenotypic characteristics of crops. According to the growth characteristics of crops (such as leaf area index, plant height, root development, etc.), the boundary conditions of the growth response stage are set. Environmental factors (such as temperature, humidity, light, etc.) are introduced to optimize the division of growth stages. The collected phenotypic data is denoised, standardized and processed for missing values to ensure the quality of the data. Key phenotypic characteristics corresponding to the crop growth stage are extracted to construct a feature set, and factors such as seasonality and environmental impact are considered. Deep learning (such as LSTM, convolutional neural network) or traditional machine learning methods (such as random forest, support vector machine) are used to establish a model to predict the growth trend of crops. Methods such as cross-validation and error analysis are used to evaluate the accuracy of the model, and adjustments and optimizations are made according to the actual situation.

[0031] Step S2: Input the phenotypic feature data into the growth trend prediction model, and simulate the crop growth process through the digital twin engine to generate growth trend evolution data; extract the physiological abnormality characteristics of the growth trend evolution data to generate stress response signals; In the embodiments of the present invention, the phenotypic data collected in step S1 (such as leaf color, plant height, leaf area index, etc.) is input into the growth trend prediction model in the form of time series or state vector. According to the requirements of the prediction model, the input data is appropriately normalized to ensure that the phenotypic characteristic data at different time points can be compared and calculated on the same scale. A suitable growth prediction model is selected (such as a regression model based on machine learning, a long short-term memory network LSTM of deep learning, etc.) to predict the growth trend of crops based on the phenotypic data. Based on the physical and biological models of crop growth (such as crop growth models, climate models, etc.), a virtual simulation environment for crop growth is constructed through a digital twin engine. Digital twin technology can establish a virtual crop growth model and input real-time phenotypic characteristic data and environmental data into the model. Using the real-time collected data, such as temperature, humidity, light, soil conditions, etc., the growth process of crops under different environmental conditions is simulated in real time through the digital twin engine. These simulation processes are based on the output of the growth trend prediction model to generate crop growth evolution data. The digital twin engine will generate detailed crop growth evolution data, including the phenotypic characteristics of crops at different growth stages, ecological environment changes, and the evolution of crop growth potential. According to the known physiological abnormality patterns (such as water stress, pests and diseases, etc.), the growth evolution data of crops is subjected to anomaly detection. Common physiological abnormalities include leaf atrophy, color change, plant height growth stagnation, etc. Key features (such as growth rate, leaf area index, root growth rate, etc.) are extracted from the simulated growth trend evolution data to analyze whether there are abnormal fluctuations inconsistent with the normal growth mode. Statistical analysis methods (such as Z-score, outlier detection algorithm, rate of change calculation, etc.) are used to quantitatively analyze the abnormal fluctuations in the crop growth process and extract physiological abnormality signals. According to the known stress patterns of crop growth, machine learning or deep learning models (such as support vector machines, convolutional neural network CNN, etc.) are used to classify the extracted physiological abnormality characteristics to identify the stress types. Pattern recognition and classification are performed on the extracted abnormal characteristics to match the abnormal growth mode of the crop with the known stress response mode, so as to identify whether the crop is under environmental stress and determine the type of stress (such as drought, pests and diseases, etc.). Time series analysis is performed on the growth data of the crop to observe the change trends before and after the stress event to further confirm the stress response of the crop.

[0032] Step S3: Obtain the three-dimensional phenotypic map of the canopy based on the stress response signal; locate the vascular bundle nodes of the three-dimensional phenotypic map of the canopy through morphological topology analysis; obtain the phloem microscopic image based on the node coordinates and perform chloroplast fluid state detection. If it is detected that the fluid state entropy value exceeds the preset threshold range, it is determined as the phenomenon of photosynthetic assimilation blockage; In the embodiments of the present invention, by using multi-source sensors (such as hyperspectral imaging, light detection and ranging (LiDAR), structured light sensors, etc.) to perform an omni-directional scan on the crop canopy, three-dimensional point cloud data of the canopy is obtained. The three-dimensional model of the canopy is constructed using the point cloud data. By adopting computer vision or geometric reconstruction algorithms, the point cloud data is converted into a complete three-dimensional phenotypic map, which can reflect the morphology, distribution of the canopy, and the spatial relationship of the structures of each part. The relevant data extracted from the three-dimensional phenotypic map of the canopy is combined with information such as crop growth and environmental stress to provide a basis for subsequent analysis. Morphological processing methods (such as edge detection, region growing, watershed algorithm, etc.) are used to analyze the three-dimensional phenotypic map, and each key node in the canopy structure is extracted. Based on topological principles (such as node connectivity, grid segmentation, etc.), the distribution of the vascular bundle system in three-dimensional space is analyzed. In particular, in-depth analysis is carried out on the connection relationship, distribution density, etc. of the vascular bundles to identify the spatial positions of the vascular bundle nodes. Graphic processing technology is used to accurately locate the positions of the vascular bundle nodes in the three-dimensional map. The nodes are marked using a three-dimensional coordinate system, and their precise spatial position coordinates are obtained. According to the coordinates of the vascular bundle nodes located in the previous step, the specific position of the phloem is obtained. A microscopic imaging device (such as a confocal microscope, a laser scanning microscope, etc.) is used to perform high-resolution imaging on this position. The phloem tissue is imaged under the microscope to obtain an image containing details such as chloroplasts and cell walls. Special attention is paid to capturing the arrangement and flow state of the chloroplasts to determine whether photosynthesis is proceeding normally. The chloroplast flow in the microscopic image is analyzed, and image processing algorithms (such as particle tracking, optical flow method, etc.) are used to detect the flow condition of the chloroplasts. According to the flow characteristics of the chloroplasts, it is judged whether there is an abnormal flow state. According to the flow pattern of the chloroplasts, the entropy value of the flow state is calculated to quantify the degree of disorder of the flow. The higher the entropy value, the more disordered the flow, and the more likely it is that the photosynthesis is blocked. The information entropy calculation method, such as Shannon entropy, is used to analyze the flow state data. According to the experimental data or previous research, a preset threshold interval of the entropy value is set. Usually, when the entropy value is within the normal range, it indicates that the photosynthetic process is normal, and when the entropy value exceeds the threshold interval, it is a signal of photosynthesis blockage. If it is detected that the entropy value of the chloroplast flow state exceeds the preset threshold interval, it is determined that there is a blockage in the photoassimilation process, specifically due to the blockage of photosynthesis caused by stress (such as water stress, nutrient deficiency, pests and diseases, etc.).

[0033] Step S4: Analyze the regulatory defects of the photoassimilation blockage phenomenon, and perform parameter self-correction on the growth trend prediction model through the regulatory defects to generate an optimized growth decision model.

[0034] In the embodiments of the present invention, based on the photosynthetic assimilation blockage phenomenon identified in the previous step, the regulation defects are analyzed. Common regulation defects include decreased stomatal conductance, damaged chloroplasts, decreased photosystem II function, etc. Through environmental monitoring data (such as temperature, humidity, light, CO 2 concentration, etc.) and crop growth data (such as nutritional status, soil humidity, etc.), combined with crop physiology theory, the influence of stress factors on the photosynthetic process is deeply analyzed. Combining molecular biology data (such as gene expression levels, enzyme activities, chlorophyll content, etc.), the internal physiological mechanism leading to photosynthetic assimilation blockage is analyzed, and relevant regulation defects are identified. According to the analysis results, a regulation defect model reflecting photosynthetic assimilation blockage is established. This model can be a mathematical model or a machine learning-based prediction model, reflecting the reasons for the blocked photosynthesis of crops under stress conditions. Determine the important parameters affecting photosynthetic efficiency and crop growth, usually including photosynthetic rate, stomatal conductance, chloroplast health status, photosynthesis-related enzyme activities, etc. Combining the existing parameters in the growth trend prediction model, analyze the influence of regulation defects, and adjust or introduce new model parameters to reflect the influence of these defects. For example, if the stomatal conductance decreases, the relevant parameters of stomatal conductance in the model can be adjusted accordingly, or the photosynthetic rate can be corrected. Through self-correction algorithms (such as Kalman filtering, Bayesian inference, etc.), the information of regulation defects is incorporated into the update process of model parameters, enabling the model to adaptively adjust under different growth stages and environmental conditions. Iteratively adjust multiple times in the actual environment, and continuously optimize and correct the model parameters using real-time collected data to ensure that the growth trend prediction model can accurately reflect the growth trend of crops under stress conditions. Determine the key decision variables for optimizing the growth decision model, such as management measures like irrigation, fertilization, pest and disease control, and their adjustment strategies under different environments and stress conditions. Based on optimization theories (such as genetic algorithms, particle swarm optimization, simulated annealing, etc.), design an optimized growth decision model. This model automatically recommends the optimal crop management measures by adjusting the weights and thresholds of decision variables to cope with different stress situations. For multiple objectives in the crop growth process (such as maximizing photosynthetic rate, increasing crop yield, minimizing resource waste, etc.), apply multi-objective optimization methods for comprehensive optimization. Based on the optimization results, provide accurate growth management decision suggestions for agricultural managers, such as when to perform operations like irrigation, fertilization, and light regulation to promote the healthy growth of crops under stress conditions.

[0035] Preferably, step S1 includes the following steps: Step S11: Collect crop phenotypic characteristic data through a multi-source sensing array, where the crop phenotypic characteristic data includes canopy spectral reflection data and rhizosphere microenvironment data; Step S12: Calculate the NDVI index for the canopy spectral reflection data to obtain crop growth status indicators; Step S13: Extract the moisture and temperature of the rhizosphere microenvironment data, and analyze the root prosperity of the crop in combination with the crop growth status indicators to generate crop root prosperity data; Step S14: Divide the crop growth response stage based on the crop root prosperity data to obtain the crop growth response stage; Step S15: Construct a growth trend prediction model according to the crop growth response stage.

[0036] In the embodiment of the present invention, by using a variety of sensing devices such as hyperspectral sensors, temperature and humidity sensors, soil moisture sensors, and infrared sensors, the crop phenotypic characteristic data is jointly collected. The sensor array can be installed above the crop growth area, in the field, and around the rhizosphere. The reflected spectral data of the crop canopy is obtained by the hyperspectral remote sensing instrument in different bands, covering visible light, near-infrared, short-wave infrared and other bands. The soil temperature, humidity and other rhizosphere microenvironment parameters are obtained by devices such as soil moisture sensors and temperature and humidity detectors. Using the spectral reflection data of the canopy, the NDVI index is calculated according to the following formula: Where, is the reflection value in the near-infrared band, It is the reflection value in the red light band. The collected spectral data is preprocessed, including noise removal, spectral normalization, etc., to ensure the accuracy of NDVI. The greenness, health status, and growth stage of crops are reflected by the value of NDVI. Generally, a high NDVI value indicates healthy crop growth, while a low value indicates problems such as water or nutrient deficiency. Soil moisture (such as soil moisture sensors) and temperature (such as temperature sensors) data are extracted from the data collected by the sensors. Rhizosphere moisture and temperature directly affect the growth and development of roots. The NDVI index is combined with rhizosphere moisture and temperature to evaluate the health status of crop roots. If the NDVI index is high and the rhizosphere moisture and temperature are appropriate, it indicates that the crop roots are prosperous; otherwise, there is a situation of insufficient water or too high temperature. According to the rhizosphere microenvironment data and crop growth status indicators, machine learning algorithms or mathematical models are used to calculate the root prosperity data. The model can consider multiple factors such as water, temperature, and root activity to judge the prosperity degree of roots. According to the crop growth law, the growth process of crops is divided into several response stages. Common growth stages include the germination stage, vegetative growth stage, flowering stage, fruiting stage, maturity stage, etc. Based on the root prosperity data, different crop growth response stages are divided by setting thresholds or model judgment. Generally, the stage with a relatively high root prosperity and good growth status corresponds to the vegetative growth stage, while a low root prosperity indicates entering a stage of growth retardation or decline. As the growth season progresses, by continuously monitoring the change of root prosperity, the crop growth response stage is dynamically adjusted, and data support is provided for subsequent prediction modeling. According to the dynamic characteristics of crop growth, an appropriate growth prediction model is selected, such as a regression model based on machine learning (support vector machine, decision tree, neural network, etc.), or a growth model based on physiology (such as a plant growth simulation model). Based on the crop growth response stage data obtained in step S14, combined with the root prosperity, NDVI data, and other growth environment factors (such as temperature, light, water, etc.), a multiple regression or deep learning model is constructed to predict the future growth trend of the crop. Historical data is used for training, and the cross-validation method is used to verify the accuracy of the model. The model can be parameter-tuned according to the actual situation to ensure good prediction ability under different growth stages and environmental conditions. After the model training is completed, real-time collected phenotypic data is input to predict the growth status of the crop in the next period of time, including growth trend, pest and disease risk, yield estimation, etc.

[0037] Preferably, step S15 includes the following steps: Step S151: Collect historical crop stage-matched growth data based on the crop growth response stage; Step S152: Divide the historical crop stage-matched growth data into data sets to generate a model training set and a model test set; Step S153: Perform model training on the model training set through the support vector machine algorithm to generate a preliminary growth trend prediction model; Step S154: Use the model test set to perform model optimization iteration on the preliminary growth trend prediction model, thereby generating a growth trend prediction model.

[0038] In the embodiments of the present invention, by collecting historical growth data of different crop varieties at different growth stages, these data should include crop growth indicators (such as plant height, leaf area, root weight, etc.), environmental data (such as air temperature, precipitation, light intensity, etc.), and growth results (such as yield, fruit size, etc.). According to the crop growth response stages divided in step S14, select the historical data corresponding to the current crop growth response stage. For example, if the current crop is in the vegetative growth stage, select the data of historical crops in the same stage for matching. To ensure the comparability between different historical data sets, perform standardization processing on each item of data, such as normalization, unit conversion, etc., to adapt to subsequent model training. According to the growth data matched by historical crop stages, divide the data set randomly or according to time series. Usually, 70%-80% of the data is used for the training set, and 20%-30% of the data is used for the test set. The cross-validation method can be used to further optimize the data division. Extract features related to the crop growth trend from the historical growth data, such as meteorological data, soil parameters, crop physical characteristics (chlorophyll content, photosynthesis rate, etc.), and the historical growth trajectory of the crop. Ensuring that the selected features are crucial for the accuracy of the prediction model. Clean the data, remove missing values, outliers, and process imbalanced data (such as oversampling or undersampling) to ensure the quality of the training set and the test set. Select the support vector machine (SVR) algorithm suitable for regression problems, which is especially suitable for processing small samples and high-dimensional data and has good classification and regression effects. Use methods such as grid search or random search to tune the hyperparameters (such as kernel function type, penalty parameter C, regression accuracy parameter, etc.) of the SVM model. Commonly used kernel functions include linear kernel, radial basis kernel (RBF), etc. Use the divided training set to train the SVM model, input the feature data of the crop, and output the prediction results of the crop growth trend (such as the growth state in the next few days or weeks, yield estimation, etc.). After training, obtain a pre-model for predicting the growth trend. This model can well fit the crop growth data on the training set and has a certain prediction ability. Use the divided test set to test the pre-model for predicting the growth trend and evaluate its performance on the test set. Commonly used evaluation indicators include mean square error (MSE), mean absolute error (MAE), coefficient of determination (R²), etc. Evaluate the performance of the model and find problems such as overfitting, underfitting, or excessive prediction errors. The model can be optimized by adjusting hyperparameters, increasing the sample size, optimizing feature selection, etc. During the optimization process, combined with the performance feedback of the model, repeatedly adjust the model parameters, iterate the training, and gradually improve the prediction accuracy and generalization ability of the model. After completing the optimization iteration, generate the final growth trend prediction model and perform the final verification through the validation set to ensure its reliability on unknown data.

[0039] As an example of the present invention, refer to Figure 2As shown, in this example, step S2 includes: Step S21: Input the phenotypic characteristic data into the growth trend prediction model to predict the crop growth trend and generate crop growth trend prediction data; Step S22: Simulate the crop growth process for the growth trend prediction data through the digital twin engine to generate growth trend evolution data; Step S23: Extract the physiological abnormality characteristics of the growth trend evolution data and perform time series analysis on the physiological abnormality characteristics to generate a physiological abnormality time window; Step S24: Quantify the stress response index for the growth trend evolution data according to the physiological abnormality time window to generate a stress response index; issue a signal warning for the physiological abnormality characteristics through the stress response index to generate a stress response signal.

[0040] In the embodiments of the present invention, by preprocessing the input phenotypic feature data, including data standardization, denoising, missing value filling, etc., it is ensured that the data is suitable for model prediction. The processed phenotypic feature data is input into the growth trend prediction model to generate growth trend prediction data of the crop at different time points. This data usually includes indicators such as the growth rate of the crop, plant height, leaf area index, root development, photosynthesis, etc. The generated growth trend prediction data of the crop can be used to predict the growth trend of the crop and the environmental pressures faced in the next few days or weeks. Based on the physical, environmental, ecological and other parameters of crop growth, a digital twin model is constructed to simulate the growth process of the crop under different environmental conditions. Using the growth trend prediction data as the initial input, the digital twin engine performs dynamic simulation according to the physical model of crop growth and environmental variables to generate growth evolution data of the crop at different future time points. The digital twin simulation results can include the growth trajectory of the crop, development process, health status, growth response under environmental changes, etc. This data provides a time-series change in growth status for subsequent analysis. Extract features indicating physiological abnormalities from the simulated growth trend evolution data, such as changes in chlorophyll content, photosynthetic rate, stem thickness, leaf area index, etc. If some indicators change too fast or exceed the normal range, it is a signal of stress response. Use time series analysis methods (such as autoregressive integrated moving average (ARIMA), sliding window analysis, Fourier transform, etc.) to analyze the extracted physiological features. By performing trend analysis and periodic analysis on the time series of these features, identify the moments of abnormal changes. According to the results of time series analysis, determine the time period (physiological abnormal time window) during which physiological abnormalities occur in the crop growth process. These windows represent the critical periods of stress or abnormal conditions. Calculate the stress response index by quantifying the key physiological features (such as photosynthesis rate, chlorophyll content, plant growth rate, etc.) of the crop within the physiological abnormal time window. Common methods are based on the normalized difference (for example, NDVI) or an integrated scoring model based on environmental factors (such as the weighted average of the influence of temperature, humidity, etc.). Set the threshold of the stress response index according to the crop type and environmental characteristics. If the index exceeds the threshold, it indicates that the crop is in a stressed state. This threshold is usually set according to historical data or expert experience. If the stress response index exceeds the preset threshold, send out a warning signal of physiological abnormality, indicating that the crop has encountered stress (such as drought, low temperature, nutrient deficiency, etc.) within this time window. The signal can be transmitted to the farm manager or decision-making system through a graphical interface, text message, email, etc. Based on the change of the stress response index, generate a stress response signal as the basis for further intervention measures. This signal is usually combined with the implementation of agronomic measures such as crop watering, fertilization, pest and disease control to minimize the impact of stress on crop growth.

[0041] Particularly importantly, step S22 further includes the following steps: Step S221: Perform time-series dynamic pattern recognition on the growth trend prediction data to generate time-evolution sequence data; Step S222: Based on the time-evolution sequence data, perform adaptive spatial interpolation and regional equalization processing on the growth trend prediction data to generate multi-scale spatial distribution data; Step S223: Perform co-modeling of environmental variables on the multi-scale spatial distribution data to generate comprehensive environmental impact data; Step S224: Through the comprehensive environmental impact data, perform a simulation of the impact of crop growth trends on the growth trend prediction data to generate growth trend evolution data.

[0042] In the embodiments of the present invention, growth trend prediction data obtained from different time nodes is collected. This data may include meteorological factors (such as temperature, humidity, precipitation), crop growth indicators (such as chlorophyll content, plant height, etc.), and soil conditions (such as moisture, pH value, nutrients, etc.). The collected growth trend data is cleaned to remove missing values and outliers, and the data is smoothed to ensure the stationarity and usability of the data. Time series data analysis techniques, such as dynamic time warping (DTW) or long short-term memory network (LSTM) models, are used to perform pattern recognition on the growth trend prediction data. These models can identify the temporal dynamic patterns in the data and help capture the laws of the growth trend changing over time. The data is decomposed into time series to separate factors such as long-term trends, seasonal variations, and periodic fluctuations, and time evolution sequence data is generated. The results of the temporal dynamic pattern recognition are used as the time evolution sequence data to reflect the growth trend changes at different time nodes. Adaptive spatial interpolation methods, such as Kriging, inverse distance weighted interpolation (IDW), or spline interpolation, are used to infer the growth trend prediction data from known spatial positions to unknown positions. Based on the time evolution sequence data, the spatial changes in each region are calculated at different time nodes to ensure that the interpolation results can reflect the spatial heterogeneity of crop growth. Spatial interpolation is performed on each moment in the growth trend prediction data to generate spatial distribution data at each time point. The interpolation results include the continuous distribution of crop growth parameters (such as chlorophyll content, plant height, etc.) in space. Through the interpolation process, blank areas or data-sparse areas in the original data are filled to generate dense spatial data points. The generated spatial data is subjected to regional equalization processing to eliminate spatial biases between different regions. This can be achieved by introducing a regional weighted equalization algorithm to balance the spatial data, making the distribution of crop growth indicators more uniform within different geographical regions. Methods such as weighted average and regional centroid method are used to ensure the balance and consistency of the multi-scale spatial distribution data. Based on the adaptive spatial interpolation and regional equalization processing, spatial distribution data containing different scales (such as fine-grained regions, global regions, etc.) is generated. These data provide a spatial basis for subsequent co-modeling of environmental variables. According to the characteristics of crop growth, environmental variables affecting crop growth are selected, such as climate data (temperature, humidity, precipitation, etc.), soil data (pH value, soil moisture, nutrients, etc.), and light intensity. Machine learning algorithms such as multiple linear regression models, support vector machines (SVM), or random forests (RF) are used to combine the multi-scale spatial distribution data with the selected environmental variables for co-modeling of environmental variables. During the modeling process, the mutual influence between environmental variables and their comprehensive effects on crop growth are considered to generate comprehensive environmental impact data. Based on the results of the co-modeling of environmental variables, a simulation model for the impact of crop growth trends is constructed.Commonly used models include crop growth models (such as CERES, APSIM, etc.), which can take into account factors such as temperature, light, moisture, soil quality, etc. Comprehensive environmental impact data is introduced into the model, and by adjusting the model parameters, the dynamic impact of environmental factors on crop growth is simulated. Using the above models, the growth trend prediction data is simulated to generate growth trend evolution data for different time periods. The simulation process takes into account the growth processes of crops such as photosynthesis, respiration, and nutrient absorption, as well as the impact of environmental factors on these processes.

[0043] Preferably, the steps for extracting the physiological abnormality characteristics of the growth trend evolution data in step S23 include the following steps: When any of the following situations occurs, the growth trend evolution data is determined to be abnormal in growth trend evolution, and growth trend evolution abnormal data is obtained: the growth rate deviates from the optimal range by more than ±10% cm / day; the root growth length is less than 10 cm or more than 50 cm; the environmental temperature deviates from the normal range of 18 - 30 °C; the light intensity is less than 500 lux or more than 2000 lux; When the following situations occur simultaneously, the growth trend evolution data is determined to be physiological dysfunction and physiological dysfunction data is obtained: the photosynthesis rate is lower than 20 μmol / m² / s for 3 consecutive days; the leaf water content is less than 40%; the pH value deviates from the range of 5.5 - 7.5; the soil nutrient concentration is less than 50% mg / kg of the normal level; When the following situations occur simultaneously, the growth trend evolution data is determined to be water absorption failure and water absorption failure data is obtained: the root water absorption rate continuously decreases by more than 20% ml / hour; the plant transpiration rate is lower than 30% g / hour of the normal level; the soil humidity deviates from the range of 40 - 60%; the plant leaves turn yellow and cannot recover in time; Integrate the growth trend evolution abnormal data, physiological dysfunction data, and water absorption failure data to obtain the physiological abnormality characteristics of the growth trend evolution data.

[0044] In the embodiments of the present invention, by comparing the actual growth rate of the crop with the optimal growth rate range (±10% cm / day). If the actual rate exceeds this range, it is recorded as an abnormal evolution of the growth trend. Using the predicted data of the crop growth trend, check the daily change of the growth rate. If the rate deviates from the standard range of ±10%, it is determined as abnormal. The normal growth range of the root system is 10 cm to 50 cm, and exceeding this range is regarded as abnormal. Extract the root length from the root growth data and determine whether it is less than 10 cm or greater than 50 cm. If the condition is met, it is marked as abnormal. The normal range of the environmental temperature is 18°C to 30°C, and exceeding this range is regarded as abnormal. According to the environmental temperature monitoring data, determine whether the temperature exceeds the normal range of 18 - 30°C. If the temperature exceeds this range, it is marked as abnormal. The normal range of the light intensity is 500 lux to 2000 lux, and the light intensity lower than or higher than this range is regarded as abnormal. According to the light intensity sensor data, determine whether it is less than 500 lux or greater than 2000 lux. If the condition is met, it is marked as abnormal. If any one or more of the above conditions are met, abnormal data of the growth trend evolution is generated. If the photosynthesis rate is lower than 20 μmol / m² / s for three consecutive days, it is determined as a photosynthesis dysfunction. Compare the photosynthesis rate data for three consecutive days. If it is lower than 20 μmol / m² / s within three days, it is marked as a physiological dysfunction. If the leaf water content is lower than 40%, it is determined as a physiological dysfunction. Check the leaf water content data. If it is lower than 40%, it is recorded as abnormal. If the soil pH value deviates from the normal range of 5.5 - 7.5, it is regarded as a dysfunction. Obtain the soil pH value and compare it with the standard range of 5.5 - 7.5. If it exceeds this range, it is marked as abnormal. If the soil nutrient concentration is lower than 50% of the normal level (unit: mg / kg), it is determined as a physiological dysfunction. Compare the soil nutrient concentration with the normal level. If it is lower than 50%, it is recorded as abnormal. If any three of the above abnormal conditions occur simultaneously, physiological dysfunction data is generated. If the root water absorption rate continuously decreases by more than 20% (unit: ml / hour), it is determined as a water absorption failure. According to the historical data of the root water absorption rate, calculate its change trend. If the decrease exceeds 20%, it is marked as a failure. If the transpiration rate of the plant is lower than 30% of the normal level (unit: g / hour), it is determined as a water absorption failure. Compare the normal transpiration rate with the current transpiration rate. If it is lower than the 30% standard value, it is recorded as a failure. If the soil humidity deviates from the range of 40% - 60%, it is regarded as a water absorption failure. Obtain the soil humidity data. If its value deviates from the normal range of 40% - 60%, it is marked as a failure. If the plant leaves turn yellow and cannot recover to the physiological state, it is determined as a water absorption failure. According to the monitoring data of the plant leaf color change, if the leaves turn yellow and cannot recover, it is recorded as a failure. If the above conditions occur simultaneously, water absorption failure data is generated.Merge the data of abnormal growth trend evolution, physiological function disorder, and water absorption failure to form a complete dataset of physiological abnormal characteristics. Standardize these data to unify different types of abnormal markers, facilitating subsequent signal warning and decision support. Identify the integrated data, such as "abnormal growth trend", "physiological function disorder", "water absorption failure", etc., so as to provide it to the decision-making system or agronomists in subsequent links for timely adjustment of agricultural management measures.

[0045] Preferably, the obtaining of the three-dimensional canopy phenotypic map based on the stress response signal in step S3 includes: Locate the regional position coordinates of the crop canopy based on the stress response signal; Use a spectral lidar array to scan the regional position coordinates of the crop canopy to obtain crop canopy point cloud data and reflectivity; Extract the geometric topology of the crop canopy point cloud data, and set the voxel size to 2 cm³ to construct a voxel space, obtaining crop canopy voxel modeling data; Screen the abnormal reflection areas of the crop canopy voxel modeling data according to the reflectivity, and perform chlorophyll content inversion on the abnormal reflection areas to generate a three-dimensional canopy phenotypic map.

[0046] In the embodiments of the present invention, spatial positioning of the crop canopy area is carried out by utilizing stress response signals. Stress response signals (such as changes in light intensity, abnormal temperature, etc.) will generate different responses in different parts of the crop, and the regional position of the crop canopy is determined through the changes in these signals. According to the characteristics of the stress response signals, combined with the growth pattern of the crop and environmental parameters, a three-dimensional positioning algorithm (for example, triangulation based on sensor data) is used to accurately determine the regional position coordinates of the canopy. A spectral lidar array (LiDAR) is used to scan at the determined regional position coordinates of the crop canopy. The spectral lidar can obtain the three-dimensional point cloud data of the crop canopy by emitting laser beams and receiving their reflected signals. During the scanning process, the lidar will obtain the three-dimensional spatial coordinates of each point and the reflectivity data of that point. The reflectivity reflects the characteristics of the substances on the canopy surface (such as the spectral characteristics of leaves), and can be used to analyze the growth status and health of the crop. Geometric topology analysis is performed on the three-dimensional point cloud data obtained from the lidar to extract the spatial structure information of the canopy. This process includes denoising, data reduction, and analysis of the connectivity of the canopy point cloud to extract the geometric morphology of the crop canopy. The processed point cloud data is converted into voxel data. A voxel is the smallest unit in three-dimensional space, and a three-dimensional grid model can be constructed by defining a standard voxel size (such as 2 cm³). Each voxel corresponds to a spatial unit in the point cloud, thus forming a crop canopy model with spatial resolution. Reflectivity analysis is performed on each voxel in the voxel modeling data of the crop canopy. By comparing the reflectivities of different voxels, regions that deviate significantly from the normal range are identified. Abnormal regions are screened according to a preset reflectivity threshold (such as low reflectance or high reflectance regions), and these abnormal reflectance regions correspond to areas of disease, uneven light, damage, or stress in the canopy. Based on the known correlation between spectral reflectivity and chlorophyll content (such as using vegetation indices or other spectral indices), chlorophyll content inversion is performed on the screened abnormal reflectance regions. The chlorophyll content inversion model is usually based on a physical model or a statistical regression model, such as a model based on chlorophyll absorption characteristics. The inversed chlorophyll content is combined with the three-dimensional point cloud data to form a three-dimensional canopy phenotypic map. This map can display the chlorophyll distribution, health status, and existing disease or stress regions of the crop. The physiological state of the entire canopy is visualized through color coding (for example, regions with high chlorophyll content are green, and regions with low chlorophyll content are yellow or red).

[0047] Particularly importantly, the chlorophyll content inversion for the abnormal reflection region further includes: Performing polarization interference complex amplitude imaging on the abnormal reflection region to generate layered structure data of mesophyll cells; Performing fractional-order differential enhancement on the layered structure data of mesophyll cells to generate enhanced layered structure data of mesophyll cells; Extract the weak reflection signal features of the mesophyll cell layered structure enhancement data, and perform tensor-radiative transfer coupled inversion to generate the chlorophyll distribution inversion data; Verify the vortex phase topology optimization of the chlorophyll distribution inversion data, so as to generate the three-dimensional canopy phenotype map.

[0048] In the embodiments of the present invention, a high-precision lidar (such as LiDAR) or spectral imaging technology is used to locate the regions with abnormal reflection characteristics in the crop canopy. These regions typically exhibit reflectance values significantly higher or lower than the normal range. Based on the outliers in the reflectance data, the regions with abnormal reflectance are marked. A polarization interference imaging system is configured to perform high-resolution imaging on these abnormal reflection regions through specific spectral bands (for example, the red or near-infrared band). Utilizing the polarization characteristics of the system, the reflection characteristics related to the layered structure of mesophyll cells are captured, such as the reflection, refraction, and scattering information on the leaf surface. Complex amplitude imaging data containing reflection intensity and phase information is generated to describe the hierarchical distribution of cell structures. Through image processing techniques, combined with the polarization interference complex amplitude imaging results, the microscopic layered structure of mesophyll cells is reconstructed. In this step, an algorithm is applied to combine the interference effect of light with the microstructure of cells to obtain detailed cell tissue hierarchical data, such as cell wall thickness, mesophyll cell arrangement, etc. Based on the fractional-order differential theory, an appropriate order parameter (for example, 0.5 order or 0.8 order) is selected to enhance the reconstructed mesophyll cell layered structure data. This process can effectively enhance the detail features of cell structures in the image, such as the contour of the cell wall and the changes in cell gaps. Through fractional-order differential processing, the edges and transition regions of mesophyll cells are highlighted, the contrast of the image is improved, and at the same time, the influence of low-frequency noise is reduced. In particular, the low-reflection signal regions (such as cell gaps, cavity regions, etc.) are enhanced, making the changes in chlorophyll distribution more obvious in the image. The enhanced data is applied to the subsequent chlorophyll distribution inversion process. The data after fractional-order differential enhancement has clearer structural features, facilitating subsequent calculations. From the enhanced mesophyll cell layered structure data, threshold segmentation technology is used to extract the regions with weak reflection intensity, which are usually closely related to the chlorophyll distribution. For the weak reflection regions, Gaussian filtering or other smoothing processing techniques are used to remove noise to ensure the accuracy of the extracted signal features. Based on the extracted weak reflection signals, a mathematical model is established to describe the relationship between reflectance and absorption coefficient. This model takes into account the influence of chlorophyll distribution to correlate the reflectance data with chlorophyll concentration. The tensor-radiative transfer coupled inversion algorithm is used, which is an inversion technique that combines the radiative transfer model with chlorophyll distribution. This method infers the chlorophyll concentration based on the known optical properties and reflection characteristics. Using this algorithm, a set of chlorophyll distribution inversion data is generated, accurately reflecting the chlorophyll concentration in different regions of the canopy. Vortex phase analysis is performed on the chlorophyll distribution inversion data to detect the phase characteristics in the data. This analysis can reveal the spatial topological characteristics of chlorophyll distribution, such as the distribution of chlorophyll at different levels, regions, or plant parts. Vortex phase is mainly used to identify the vortex-type changes or complex topological forms in the leaf tissue structure. Based on the vortex phase analysis, the chlorophyll distribution data is optimized to process the topological irregularities in the data.Using an optimization algorithm, adjust the phase information of the data to ensure that the inversion results are consistent with the actual observed data. Adopt the minimum error algorithm to optimize the spatial distribution of the chlorophyll inversion data and eliminate the influence caused by instrument noise or calculation errors. Compare the optimized chlorophyll distribution data with the actual chlorophyll concentration measurement values to verify the accuracy of the inversion model. If the verification result is not satisfactory, further adjust the model parameters and repeat the inversion and optimization process until the preset accuracy requirement is met. Combine the optimization results of the chlorophyll distribution to generate the final three-dimensional canopy phenotypic map. This map will contain detailed chlorophyll concentration data and be able to comprehensively display the growth of the crop canopy. Through three-dimensional modeling technology, finally generate a visual three-dimensional canopy phenotypic map for subsequent crop growth analysis and decision support.

[0049] Preferably, the positioning of the vascular bundle nodes of the three-dimensional canopy phenotypic map through morphological topology analysis in step S3 includes: Extract the local geometric features of the three-dimensional canopy phenotypic map to obtain curvature, normal, and neighborhood relationships; Perform density clustering on the three-dimensional canopy phenotypic map according to the curvature, normal, and neighborhood relationships to generate feature clusters; Use the feature clusters to screen the candidate vascular bundle regions in the three-dimensional canopy phenotypic map and perform connectivity analysis on the candidate vascular bundle regions to obtain regional connectivity; Perform branch structure analysis on the candidate vascular bundle regions to obtain regional branch data; Based on the regional connectivity and regional branch data, perform vascular bundle morphological topology screening on the candidate vascular bundle regions to obtain the vascular bundle nodes of the three-dimensional canopy phenotypic map.

[0050] In the embodiments of the present invention, local curvature analysis is performed on each voxel or point in the three-dimensional canopy phenotypic map. Curvature measures the degree of surface bending and is crucial for detecting morphological changes in canopy leaves, leaf edges, and concave parts. Commonly used calculation methods include Gaussian curvature and mean curvature. The normal vector of each voxel or point is calculated, and the normal vector indicates the direction of the surface at that point. The direction and magnitude of the normal reflect the local characteristics of the canopy surface morphology. The neighborhood voxels or points around each voxel are analyzed to construct an adjacency relationship. The neighborhood relationship can help identify the geometric structure and connectivity of adjacent regions, especially in identifying connected regions and patterns in the canopy structure. The density clustering algorithm (such as DBSCAN) is used to cluster the point cloud data of the canopy. This algorithm can identify local structures or regions based on the distance and density relationships between points. For example, points with similar curvatures and normal directions are grouped into the same class through clustering. The clustering results generate multiple feature clusters, and each cluster corresponds to a structural region in the canopy. These clusters can represent different tissue structures, such as leaves, branches, or vascular bundle regions. The regions containing vascular bundles are screened according to the geometric morphology of the feature clusters (such as curvature changes, normal directions, etc.). Vascular bundles usually exhibit certain morphological characteristics, such as being slender and curved. Connectivity analysis is performed on the candidate regions, and the connectivity component algorithm (such as union-find or breadth-first search) is used to identify whether there is a continuous vascular bundle structure in the three-dimensional canopy phenotypic map. Connectivity analysis can help identify whether the structure in the region is complete and continuous, thereby filtering out incomplete or scattered regions. The branch analysis algorithm (such as topological analysis or branch detection algorithm) is used to analyze the candidate vascular bundle regions. Vascular bundles usually have a typical branch structure, and branches can be identified by calculating indicators such as bifurcation points and connectivity degrees in the region. By analyzing the positions, numbers, and angles of the branch points, the branch data of the region are generated, and these branch data can be used to confirm whether it is a vascular bundle region. The branch structure of vascular bundles usually exhibits specific symmetry and biological significance. Combining the connectivity and branch data of the region, a topological screening algorithm is used to determine the final vascular bundle nodes. Vascular bundles usually have high connectivity and a specific branch structure, so these nodes can be identified through topological structure analysis. Based on the connectivity and branch structure of the screened region, the specific node positions of the vascular bundles are determined. Nodes usually refer to the starting, branching, or connection points of the vascular bundle structure, and the coordinates of these nodes have specific biological significance in the three-dimensional phenotypic map. The finally determined vascular bundle nodes will serve as important markers in the three-dimensional canopy phenotypic map, reflecting the key physiological structures and functional regions in the crop canopy.

[0051] Preferably, the obtaining of the phloem microscopic image based on the node coordinates and the detection of the chloroplast fluidity in step S3 include: Calculate the node coordinates of the vascular bundles of the canopy three-dimensional phenotypic map, and perform microscopic imaging on the vascular bundle region according to the node coordinates to obtain phloem microscopic images; Perform chloroplast detection and processing on the phloem microscopic images to generate chloroplast fluid state region data; Extract the fluid state characteristics of the chloroplast fluid state region data, and calculate the information entropy of the chloroplast fluid state region data to obtain the fluid state entropy value; Compare the fluid state entropy value with a preset threshold interval. When the fluid state entropy value is greater than the preset threshold interval, the fluid state entropy value is marked as the phenomenon of photosynthetic assimilation blockage.

[0052] In the embodiments of the present invention, according to the node data obtained by performing morphological topology analysis on the vascular bundles of the canopy three-dimensional phenotypic map in the foregoing steps, the three-dimensional coordinates of each vascular bundle node are extracted, and these node coordinates will be used for the shooting and positioning of subsequent microscopic images. According to the node coordinates, use a microscope or other imaging device to accurately shoot the vascular bundle region to obtain high-resolution phloem microscopic images. The microscopic imaging should focus on the phloem region in the vascular bundle, especially the distribution and activity of chloroplasts during the photosynthesis process. Use image processing algorithms (such as threshold segmentation, edge detection, morphological operations, etc.) to process the phloem microscopic images to extract the regions of chloroplasts. Chloroplasts usually appear as regions with specific colors or morphologies in microscopic images. Especially during the fluid state process, the movement or flow direction of chloroplasts affects their morphological characteristics. Through image processing, the active regions and states of chloroplasts are detected, and their fluid state regions are calibrated. This data includes information such as the position, morphology, and density of chloroplasts in microscopic images. According to the morphological and movement characteristics of the chloroplast fluid state region, extract the relevant characteristics of the fluid state, such as the distribution density, flow direction, and speed of chloroplasts. The changes in the fluid state can be characterized by calculating the spatial distribution pattern, shape symmetry, etc. of the chloroplast fluid state region. Use information entropy (Shannon entropy) to calculate the complexity of the chloroplast fluid state region data. Information entropy reflects the uncertainty or chaos degree of the data. A higher entropy value usually indicates a more complex or disordered state of the system. The entropy value calculation formula is: Among them, is the probability of each state in the chloroplast fluid state region, is the number of states. This step helps to quantify the regularity or disorder degree of chloroplast flow. Through experimental data or literature research, a preset threshold range is set. This range is usually set according to the flow state change range of chloroplasts under normal photosynthesis. When the flow state entropy value exceeds this range, it indicates that abnormal phenomena occur during the photosynthesis process. Compare the calculated flow state entropy value with the preset threshold range. If the flow state entropy value is greater than the threshold, it means that the flow state of chloroplasts becomes more chaotic, and there is a phenomenon of photosynthetic assimilation blockage. When the flow state entropy value exceeds the threshold range, mark it as "photosynthetic assimilation blockage phenomenon" and perform corresponding processing or early warning.

[0053] As an example of the present invention, refer to Figure 3 shown, in this example, the step S4 includes: Step S41: Collect carbon flux deficit data of blockage events based on the photosynthetic assimilation blockage phenomenon; Step S42: Calculate the elasticity coefficient of the rate-limiting enzyme of the carbon flux deficit data through metabolic control analysis; Step S43: Determine the expression levels of Rubisco activase and SPS protein, and construct an enzyme kinetic correction function; set the starch / sucrose output ratio constraint condition for the rate-limiting enzyme elasticity coefficient through the rate-limiting enzyme elasticity coefficient; Step S44: Use the starch / sucrose output ratio constraint condition to perform parameter self-correction on the growth trend prediction model to generate an optimized growth decision model.

[0054] In the embodiment of the present invention, based on the chloroplast flow state analysis result, the time point and area where the photosynthetic assimilation blockage occurs are determined. Through the carbon isotope labeling method or gas exchange analysis method (such as photosynthetic rate, CO 2Absorption rate, etc.), collect carbon flux data related to the arrest phenomenon. By comparing the carbon flux under normal photosynthesis with the carbon flux during the arrest event, carbon flux deficit data is obtained. This data reflects the carbon deficiency in crops due to photosynthetic assimilation arrest, which in turn affects their growth and yield. Through metabolic control analysis, combined with carbon flux deficit data, the activity changes of rate-limiting enzymes (such as Rubisco, etc.) that affect the photosynthesis rate are evaluated. The MCA method can reveal the control role of different enzymes in the metabolic network and their impact on carbon flux. Calculate the rate-limiting enzyme elasticity coefficient, that is, the degree of response of the rate-limiting enzyme in the metabolic network to the carbon flux deficit. The elasticity coefficient can represent the relative control role of the rate-limiting enzyme in the carbon metabolic pathway. The calculation of this coefficient is usually based on the enzyme kinetics model and obtained through data fitting. Through enzyme activity determination experiments, quantify the activity of Rubisco (Ribulose-1,5-bisphosphate carboxylase / oxygenase) activase and analyze its contribution to carbon assimilation in photosynthesis. Through techniques such as Western blot or immunoprecipitation, quantitatively analyze the expression level of sucrose phosphate synthase (SPS) protein. SPS is crucial for sugar synthesis and distribution and can reflect the distribution of carbon assimilation products between starch and sucrose. Based on the expression levels of Rubisco activase and SPS protein, construct an enzyme kinetics correction function. This function can correct the reaction rate catalyzed by the enzyme to adapt to the carbon flux deficit under the arrest phenomenon and simulate the impact of photosynthetic assimilation arrest on the metabolic pathway. According to the enzyme kinetics correction function, set the output ratio constraint conditions of starch and sucrose. These constraint conditions can determine the flow direction of carbon, that is, determine the distribution ratio of carbon between starch storage and sucrose transport, through the relative activities of Rubisco and SPS. Use the starch / sucrose output ratio constraint conditions obtained in step S43 to correct the parameters of the growth trend prediction model. By adjusting the metabolic parameters in the model, make it more conform to the actual carbon metabolism and growth process. Through parameter self-correction, obtain an optimized growth decision model. This model can more accurately predict the growth status and yield of crops under the photosynthetic assimilation arrest phenomenon and provide targeted decision support, such as adjusting environmental factors such as light, temperature, or moisture, so as to effectively alleviate the negative impact brought by the arrest phenomenon.

[0055] Preferably, step S44 includes the following steps: Step S441: Judge the starch / sucrose output ratio. When the starch / sucrose output ratio is less than 0.5, generate the first constraint condition; when the starch / sucrose output ratio is between 0.5 and 1.5, generate the second constraint condition; when the starch / sucrose output ratio is greater than 1.5, generate the third constraint condition; Step S442: Perform the first type of parameter self - calibration on the growth trend prediction model based on the first constraint condition to obtain the first constraint adjustment parameter, where the first type of parameter self - calibration includes increasing the photosynthetic rate parameter in the growth trend prediction model to 0.5 μmol / m² / s, increasing the starch synthesis rate to 0.2 mol / h, decreasing the sucrose transport rate to 0.05 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S443: Perform the second type of parameter self - calibration on the growth trend prediction model based on the second constraint condition to obtain the second constraint adjustment parameter, where the second type of parameter self - calibration includes maintaining the starch synthesis rate in the growth trend prediction model at 0.15 mol / h, maintaining the sucrose synthesis rate at 0.1 mol / h, keeping the sucrose transport rate at 0.05 mol / h, maintaining the sucrose consumption rate at 0.05 mol / h, and keeping the starch storage rate and sucrose utilization rate within the preset optimal range without additional adjustment; Step S444: Perform the first type of parameter self - calibration on the growth trend prediction model based on the third constraint condition to obtain the third constraint adjustment parameter, where the third type of parameter self - calibration includes decreasing the starch synthesis rate in the growth trend prediction model to 0.2 mol / h, increasing the sucrose synthesis rate to 0.15 mol / h, increasing the sucrose transport rate to 0.1 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S445: Integrate the first constraint adjustment parameter, the second constraint adjustment parameter, and the third constraint adjustment parameter into a parameter self - calibration strategy, and use the parameter self - calibration strategy to perform decision training on the growth trend prediction model to generate an optimized growth decision model.

[0056] In the embodiments of the present invention, classification is carried out according to the current starch / sucrose output ratio of the crop: when the starch / sucrose output ratio is less than 0.5, a first constraint condition is generated, indicating that there is too much starch storage and relatively less sucrose transport at this time. When the starch / sucrose output ratio is between 0.5 and 1.5, a second constraint condition is generated, indicating that the output ratio of starch and sucrose is in a relatively balanced state at this time. When the starch / sucrose output ratio is greater than 1.5, a third constraint condition is generated, indicating that there is too much sucrose transport and relatively less starch storage at this time. In the case where the starch / sucrose output ratio is less than 0.5, key parameters in the growth trend prediction model are adjusted by self-correction of the first type of parameters. The photosynthetic rate parameter is increased to 0.5 μmol / m² / s to promote carbon assimilation and make up for the carbon flux deficit. The starch synthesis rate is increased to 0.2 mol / h to promote starch storage and balance excessive sucrose synthesis. The sucrose transport rate is reduced to 0.05 mol / h to reduce the transfer of sucrose in the plant and avoid unnecessary energy consumption. The sucrose consumption rate is increased to 0.08 mol / h to enhance the utilization efficiency of sucrose. The parameters obtained according to the above adjustments are used as the first constraint adjustment parameters. In the case where the starch / sucrose output ratio is between 0.5 and 1.5, the system balance is maintained by self-correction of the second type of parameters without significant adjustment. The starch synthesis rate is maintained at 0.15 mol / h within the normal range to prevent excessive starch accumulation or insufficient synthesis. The sucrose synthesis rate is maintained at 0.1 mol / h within a reasonable range. The sucrose transport rate is maintained at 0.05 mol / h to ensure the normal transfer and transportation of sucrose. The sucrose consumption rate is maintained at 0.05 mol / h to maintain the effective utilization of sucrose. The starch storage rate and sucrose utilization rate are maintained within the preset optimal range without additional adjustment. There is no need to significantly adjust the parameters and maintain a balanced state. In the case where the starch / sucrose output ratio is greater than 1.5, the parameters in the growth trend prediction model are adjusted by self-correction of the first type of parameters to promote sucrose transport and reduce starch storage. The starch synthesis rate is reduced to 0.2 mol / h to avoid excessive starch accumulation. The sucrose synthesis rate is increased to 0.15 mol / h to increase the production of sucrose and improve carbon flow. The sucrose transport rate is increased to 0.1 mol / h to enhance the transport and distribution of sucrose in the plant. The sucrose consumption rate is increased to 0.08 mol / h to ensure the effective utilization of sucrose in the plant. The parameters obtained according to the above adjustments are used as the third constraint adjustment parameters. The first constraint adjustment parameters, the second constraint adjustment parameters, and the third constraint adjustment parameters are combined to form a complete parameter self-correction strategy. The integrated self-correction strategy is applied to the growth trend prediction model for decision-making training. Adaptive adjustment is carried out according to the current starch / sucrose output ratio and the phenomenon of photosynthetic assimilation blockage to help predict the growth status and yield of the crop.Through decision-making training, an optimized growth decision model is generated, which can automatically adjust relevant parameters under different starch / sucrose output ratios to achieve the optimization of crop growth.

[0057] Therefore, in any aspect, 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. Accordingly, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be embraced within the present invention.

[0058] The above are only specific embodiments 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 invented herein.

Claims

1. A method for analyzing and making decisions on crop growth monitoring data, characterized in that: The following steps are involved: Step S1: collecting crop phenotypic characteristic data through a multi-source sensor array; dynamically dividing crop growth response stages according to phenotypic characteristics; and constructing a growth trend prediction model based on the crop growth response stages; Step S2: input the phenotypic characteristic data into the growth trend prediction model, and simulate the crop growth process through the digital twin engine to generate growth trend evolution data; extract the physiological abnormal characteristics of the growth trend evolution data to generate a stress response signal; Step S3: obtaining a three-dimensional phenotypic map of the canopy based on the stress response signal; locating the vascular bundle nodes of the three-dimensional phenotypic map of the canopy through morphological topological analysis; obtaining a phloem microscopic image based on the node coordinates and performing chloroplast flow state detection, and if the flow state entropy value is detected to exceed a preset threshold range, it is determined to be a photosynthetic assimilation block phenomenon; Step S4: analyzing the regulation defects of the photosynthetic assimilation blockade phenomenon, and self-correcting the parameters of the growth trend prediction model through the regulation defects to generate an optimized growth decision model.

2. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting crop phenotypic characteristic data through a multi-source sensor array, wherein the crop phenotypic characteristic data includes canopy spectral reflectance data and rhizosphere microenvironment data; Step S12: Calculate the NDVI index for the canopy spectral reflectance data to obtain a crop growth status index; Step S13: extracting the moisture and temperature of the rhizosphere microenvironment data, and analyzing the root prosperity of the crop in combination with the crop growth status index to generate crop root prosperity data; Step S14: dividing the crop growth response stages based on the crop root prosperity data to obtain the crop growth response stages; Step S15: constructing a growth trend prediction model according to the crop growth response stage.

3. The method for analyzing and making decisions on crop growth monitoring data according to claim 2, characterized in that: Step S15 includes the following steps: Step S151: collecting historical crop stage matching growth data based on the crop growth response stage; Step S152: dividing the historical crop stage matching growth data into data sets to generate a model training set and a model test set; Step S153: Performing model training on the model training set by using a support vector machine algorithm to generate a growth trend prediction pre-model; Step S154: Utilize the model test set to perform model optimization iteration on the growth trend prediction pre-model, thereby generating a growth trend prediction model.

4. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: inputting the phenotypic characteristic data into a growth situation prediction model to perform crop growth situation prediction, thereby generating crop growth situation prediction data; Step S22: simulating the crop growth process of the growth trend prediction data through the digital twin engine to generate growth trend evolution data; Step S23: extracting physiological abnormality features of the growth trend evolution data, and performing time series analysis on the physiological abnormality features to generate a physiological abnormality time window; Step S24: quantifying the stress response index of the growth situation evolution data according to the physiological abnormality time window to generate a stress response index; issuing a signal warning for the physiological abnormality characteristics through the stress response index to generate a stress response signal.

5. The method for analyzing and making decisions on crop growth monitoring data according to claim 4, characterized in that: The step S23 of extracting physiological abnormality characteristics of growth status evolution data comprises the following steps: When any of the following conditions occurs, the growth state evolution data is judged as abnormal growth state evolution and abnormal growth state evolution data is obtained: the growth rate deviates from the optimal range by more than ±10% cm / day; the root growth length is less than 10 cm or greater than 50 cm; the ambient temperature deviates from the normal range of 18-30°C; the light intensity is less than 500 lux or greater than 2000 lux; When the following conditions occur at the same time, the growth state evolution data is judged as physiological dysfunction and the physiological dysfunction data is obtained: the photosynthesis rate is lower than 20μmol / m² / s for 3 consecutive days; the leaf water content is lower than 40%; the pH value deviates from the range of 5.5-7.5; the soil nutrient concentration is lower than 50% mg / kg of the normal level; When the following conditions occur at the same time, the growth trend evolution data is judged as water absorption failure and water absorption failure data is obtained: the root water absorption rate continues to decrease by more than 20% ml / hour; the plant transpiration rate is 30% g / hour lower than the normal level; the soil moisture deviates from the range of 40-60%; the plant leaves turn yellow and cannot recover in time; The abnormal growth state evolution data, physiological function disorder data and water absorption failure data are integrated to obtain the physiological abnormality characteristics of the growth state evolution data.

6. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: The step S3 of obtaining a three-dimensional phenotypic map of the canopy based on the stress response signal includes: Locate the regional position coordinates of the crop canopy based on the stress response signal; The spectral lidar array is used to scan the regional position coordinates of the crop canopy to obtain the crop canopy point cloud data and reflectivity; The geometric topological structure of the crop canopy point cloud data was extracted, and the voxel size was set to 2 cm³ for voxel space construction to obtain the crop canopy voxel modeling data; The abnormal reflection areas of the crop canopy voxel modeling data were screened according to the reflectivity, and the chlorophyll content in the abnormal reflection areas was inverted to generate a three-dimensional phenotypic map of the canopy.

7. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: The step S3 of locating the vascular bundle nodes of the canopy three-dimensional phenotypic map by morphological topological analysis includes: Extract local geometric features of the canopy 3D phenotypic map to obtain curvature, normal and neighborhood relationships; Density clustering of the three-dimensional phenotypic map of the canopy is performed based on curvature, normal and neighborhood relationships to generate feature clusters; Feature clustering was used to screen candidate vascular bundle regions in the canopy three-dimensional phenotypic map, and connectivity analysis was performed on the candidate vascular bundle regions to obtain regional connectivity. Conduct branch structure analysis on candidate vascular bundle regions to obtain regional branch data; Based on regional connectivity and regional branching data, the candidate vascular bundle regions were screened for vascular bundle morphology and topology, and the vascular bundle nodes of the canopy three-dimensional phenotypic map were obtained.

8. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: The step S3 of acquiring the phloem microscopic image based on the node coordinates and performing chloroplast flow state detection includes: Calculate the node coordinates of the vascular bundle nodes of the canopy three-dimensional phenotypic map, and take microscopic photos of the vascular bundle area according to the node coordinates to obtain the phloem microscopic image; Perform chloroplast detection processing on phloem microscopic images to generate chloroplast flow area data; Extract the flow characteristics of the chloroplast flow area data, and calculate the information entropy of the chloroplast flow area data to obtain the flow entropy value; The flow entropy value is compared with a preset threshold value interval. When the flow entropy value is greater than the preset threshold value interval, the flow entropy value is marked as photosynthetic assimilation retardation phenomenon.

9. The method for analyzing and making decisions on crop growth monitoring data according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: collecting carbon flux deficit data of the blocking event based on the photosynthetic assimilation blocking phenomenon; Step S42: calculating the rate-limiting enzyme elasticity coefficient of the carbon flux deficit data through metabolic control analysis; Step S43: determining the expression levels of Rubisco activating enzyme and SPS protein, and constructing an enzyme kinetic correction function; setting a starch / sucrose output ratio constraint condition for the rate-limiting enzyme elasticity coefficient through the rate-limiting enzyme elasticity coefficient; Step S44: using the starch / sucrose output ratio constraint condition to perform parameter self-correction on the growth trend prediction model to generate an optimized growth decision model.

10. The method for analyzing and making decisions on crop growth monitoring data according to claim 9, characterized in that: Step S44 includes the following steps: Step S441: Determine the starch / sucrose output ratio. When the starch / sucrose output ratio is less than 0.5, generate the first constraint condition; when the starch / sucrose output ratio is between 0.5 and 1.5, generate the second constraint condition; when the starch / sucrose output ratio is greater than 1.5, generate the third constraint condition; Step S442: Based on the first constraint condition, the growth situation prediction model is subjected to a first type of parameter self-correction to obtain a first constraint adjustment parameter, wherein the first type of parameter self-correction includes increasing the photosynthetic rate parameter in the growth situation prediction model to 0.5 μmol / m² / s, increasing the starch synthesis rate to 0.2 mol / h, reducing the sucrose transport rate to 0.05 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S443: Based on the second constraint condition, the second type of parameter self-correction is performed on the growth trend prediction model to obtain the second constraint adjustment parameter, wherein the second type of parameter self-correction includes maintaining the starch synthesis rate in the growth trend prediction model at 0.15 mol / h, maintaining the sucrose synthesis rate at 0.1 mol / h, maintaining the sucrose transport rate at 0.05 mol / h, maintaining the sucrose consumption rate at 0.05 mol / h, and maintaining the starch storage rate and sucrose utilization rate within the preset optimal range without additional adjustment; Step S444: performing a first type parameter self-correction on the growth trend prediction model based on the third constraint condition to obtain a third constraint adjustment parameter, wherein the third type parameter self-correction includes reducing the starch synthesis rate in the growth trend prediction model to 0.2 mol / h, increasing the sucrose synthesis rate to 0.15 mol / h, increasing the sucrose transport rate to 0.1 mol / h, and increasing the sucrose consumption rate to 0.08 mol / h; Step S445: Integrate the first constraint adjustment parameter, the second constraint adjustment parameter and the third constraint adjustment parameter into a parameter self-correction strategy, and use the parameter self-correction strategy to perform decision training on the growth trend prediction model to generate an optimized growth decision model.

Images