A crop growth trend analysis and yield evaluation method

Abstract

The application belongs to the technical field of agricultural informatization, and discloses a crop growth trend analysis and yield evaluation method, which comprises the following steps: capturing crop growth monitoring data in real time through an Internet of Things edge node cluster; the crop growth monitoring data comprises crop growth data, growth environment data and crop growth image data; an environment weight mechanism is introduced, the environment weight of each time point is calculated according to the time series growth environment data, and the adjustment coefficient of the weight influence degree is dynamically adjusted through a deviation feedback mechanism; the improved moving average algorithm is used in combination with the growth environment data to smooth the crop growth data, the window size is automatically adjusted according to the environment fluctuation, and the smoothed crop growth data is obtained; the Kalman filtering algorithm is used to filter the noise of the smoothed crop growth data, and crop growth analysis data is generated through refinement; and the healthy growth of crops is ensured, and yield loss is avoided.

Description

Technical Field

[0001] This invention relates to the field of agricultural information technology, and more specifically, to a method for analyzing crop growth trends and evaluating yield. Background Technology

[0002] Patent publication number CN114997535A discloses a smart agricultural production big data intelligent analysis method and system platform. The method includes: real-time acquisition of image data on crop growth status and the entire production process; preprocessing the real-time acquired image data; constructing and training a prediction model for crop growth status and the entire production process, including a first prediction sub-model and a second prediction sub-model; inputting the preprocessed image data into the first and second prediction sub-models respectively to obtain the first and second prediction results for crop growth status and the entire production process; and weighted summation to obtain the final prediction result for crop growth status and the entire production process. This invention constructs an agricultural big data platform based on big data and neural network technologies, which is fully adaptable to the development direction of agricultural big data and intelligent information analysis technologies.

[0003] Existing methods for analyzing crop growth status and evaluating yield have the following shortcomings:

[0004] Crop growth data fluctuates dramatically due to environmental changes and observational errors, causing models to fail to accurately reflect long-term growth trends and resulting in inaccurate predictions. Crop growth status is typically influenced by environmental conditions (such as climate, soil, and pests), and models may fail to adapt to rapid environmental changes, leading to excessive prediction errors. Ignoring the impact of environmental fluctuations on growth data means models cannot effectively adapt to changes in environmental conditions, thus limiting the system's adaptability to different types of environmental fluctuations and affecting its stability and accuracy. Crop growth data may contain significant noise caused by random fluctuations such as short-term climate changes and measurement errors, resulting in a lack of data continuity and affecting the accuracy of subsequent analyses, such as pest and disease early warning. A fixed sliding window size, coupled with frequent changes in environmental conditions, means the system lacks an adaptive mechanism. Failure to incorporate environmental weights and differential adjustment coefficients from the growth environment data can lead to underestimation or over-averaging of the importance of environmental conditions, thus affecting the model's responsiveness to environmental changes.

[0005] Considering only a single or a few environmental factors may overlook other important factors such as soil quality, air quality, and pests and diseases. Crop growth is affected by the interaction of multiple environmental factors, and ignoring these factors may lead to an incomplete assessment of crop growth and affect the accuracy of growth prediction. Different types of environmental fluctuations will have different effects on crop growth, and static smoothing strategies may not be able to adapt to these changes. Failure to adjust the sensitivity to environmental changes may result in insufficient or excessive response to certain environmental factors. Without an effective dynamic smoothing mechanism, short-term data fluctuations caused by environmental fluctuations may not be eliminated in time, leading to unstable input data for the model and affecting the accuracy of subsequent analysis.

[0006] When measurement errors are large or environmental fluctuations are drastic, models may rely too heavily on predicted or observed values, leading to inaccurate predictions of crop growth status. Without flexible adjustments, models may fail to adapt to sudden environmental changes in a timely manner, thus missing key changes in crop growth and affecting the reliability and accuracy of long-term trends.

[0007] In view of this, the present invention proposes a method for analyzing crop growth status and evaluating yield to solve the above problems. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for analyzing crop growth status and evaluating yield, comprising:

[0009] S1. Real-time capture of crop growth monitoring data through IoT edge node clusters; crop growth monitoring data includes crop growth data, growth environment data, and crop growth image data;

[0010] S2. An environmental weighting mechanism is introduced, which calculates the environmental weight at each time point based on the time series growth environment data, and dynamically adjusts the adjustment coefficient of the weight's influence through a deviation feedback mechanism; the improved moving average algorithm is used in conjunction with the growth environment data to smooth the crop growth data, and the window size is automatically adjusted according to environmental fluctuations.

[0011] The Kalman filter algorithm is used to filter out noise from the smoothed crop growth data and refine it into crop growth analytical data; the growth environment data and crop growth image data are processed to obtain growth environment variable data and growth image analytical data.

[0012] S3. The crop growth analysis data, growth environment variable data and growth image analysis data are fused to obtain the crop growth status aggregate;

[0013] S4. Input the crop growth status aggregate into the pre-trained growth status assessment model to predict the crop growth status over the next n periods.

[0014] S5. Obtain historical yield data, train a crop yield prediction model based on historical yield data and the crop growth trend over the next n periods, and predict the crop yield based on the crop yield prediction model.

[0015] S6. Based on the predicted crop yield, assess whether the crop is in its optimal growth condition.

[0016] S7. If crops are not in their optimal growth state, feedback will be sent to staff through the intelligent crop growth monitoring terminal for timely intervention.

[0017] Preferably, the crop growth data includes growth morphology data and growth function data; the growth environment data includes meteorological data, soil data, air quality data and pest and disease data; the crop growth image data includes crop vegetation index images, chlorophyll content images, plant growth height images, crop canopy coverage images and leaf area index images.

[0018] Preferably, the method for obtaining the smoothed crop growth data includes:

[0019] S31. Preset growth morphology data are: in, This refers to the growth morphology data at time point t; x q Let q be the morphological index in the growth morphology data at time point t; q is the index of the morphological index, q = 1, ..., N; N is the total number of morphological indexes.

[0020] The preset growth function data is as follows: in, For growth function data at time point t; y j Let j be the j-th functional index in the growth functional data at time point t; j is the index of the functional index, j = 1, ..., M; M is the total number of functional indices;

[0021] S32, Growth morphology data at time point t Growth function data at time point t The data is pieced together to form crop growth data, which is represented as: x t ={x1,x2,...,x q ,...,x N ,y1,y2,...,y j ,...,y M}; where x tThis refers to crop growth data at time point t;

[0022] The preset growth environment data at time point t are: in, For meteorological data at time point t; Soil data at time point t; This refers to air quality data at time point t; This refers to pest and disease data at time point t;

[0023] S33. Based on the growth environment data E at time point t t The environmental weight at each time point t is calculated using the environmental weight calculation formula; the environmental weight calculation formula is: Where, ω i (E t E represents the growth environment data at time point t. t The weight corresponding to the i-th time step; E max Data for ideal growth environment; ||E t -E max || represents the growth environment data and ideal growth environment data E corresponding to time point t. max The difference between them; ΔE t This represents the rate of change of the growth environment data from time point t-1 to time point t; α is the ratio of the control growth environment data to the ideal growth environment data E. max The coefficient for adjusting the impact of the difference between the two on the weights; β is the coefficient for adjusting the impact of the rate of change of the growth environment data on the weights; i is the index of the time step;

[0024] S34. Using the adaptive formula of the difference adjustment coefficient, the control growth environment data and the ideal growth environment data E are compared. max The adjustment coefficient α, which modifies the degree of influence of the difference on the weight, is dynamically adjusted. The adaptive formula for the difference adjustment coefficient is: Where m represents the number of types of growth environment data; denoted as , where is the rate of change of the c-th growth environment data at time t; b is a constant adjusting the influence of the rate of change on the adjustment coefficient; p is an exponential factor controlling the cumulative effect of the rate of change on the adjustment coefficient; and c is the index of the growth environment data type.

[0025] S35. Use the weighted moving average formula to smooth crop growth data; the weighted moving average formula is: Among them, WA t The data represents smoothed crop growth data at time point t; x t-a γ represents crop growth data at time point ta; γ is the sliding window size; and a is the index of the sliding window.

[0026] S36. Dynamically adjust the size γ of the sliding window using a sliding window adjustment formula. The sliding window adjustment formula is: Where, γ max ω represents the maximum value of the sliding window. avg is the average weight of the growth environment data; d is a constant that controls the effect of environmental fluctuations on the sliding window size; This is the floor function.

[0027] Preferably, the method for refining and generating analytical crop growth data by using the Kalman filter algorithm to remove noise from the smoothed crop growth data includes:

[0028] Define a crop growth state transition model to reflect the change process of crop growth state over time. Based on the crop growth state at the previous time point and the control input, predict the crop growth state at the current time point.

[0029] By updating the predicted value of crop growth status at the current time point t, the Kalman gain is calculated. The Kalman gain is used to balance the weight between the predicted value and the actual observed value of crop growth status. The error covariance matrix in the Kalman gain is also updated to reflect the uncertainty of the updated crop growth status. The crop growth data after Kalman filtering and denoising is output as the final result to obtain the crop growth analysis data.

[0030] Preferably, the method for processing growth environment data and crop growth image data to obtain growth environment variable data and growth image analysis data includes:

[0031] Density clustering algorithm is used to identify and remove outliers in crop growth monitoring data, including growth environment data and crop growth image data, to obtain growth environment variable data and growth image analysis data.

[0032] Preferably, the method for obtaining the crop growth state aggregate includes:

[0033] The crop growth analysis data, growth environment variable data, and growth image analysis data were normalized by standard deviation, and converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, to obtain normalized crop growth analysis data, growth environment variable data, and growth image analysis data. The normalized crop growth analysis data, growth environment variable data, and growth image analysis data were then weighted and fused to obtain a crop growth status aggregate.

[0034] Preferably, the training method for the growth status assessment model includes:

[0035] The dataset is divided into training, validation, and test sets for training and evaluating the model's performance. A growth status assessment model is constructed using the TensorFlow deep learning library. The growth status assessment model includes an input layer, an LSTM layer, and an output layer. The input layer of the model is used to input the aggregate of historical crop growth states. The output layer of the model is used to output the crop growth status over the next n time periods. The growth status assessment model is an LSTM model.

[0036] Define the model's loss function, using the L2 regularized mean squared error loss function to measure the difference between the model's predicted values ​​and the true values; train the growth status assessment model using the training set, update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the growth status assessment model by calculating the accuracy metric.

[0037] The Adam optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback of the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The pre-trained growth status evaluation model was used to predict the current crop growth status aggregate to obtain the crop growth status in the next n time periods.

[0038] Preferably, the historical yield data includes historical crop yield per unit area, historical total crop yield, historical crop yield under different environments, and historical yield fluctuation data.

[0039] Preferably, the training method for the crop yield prediction model includes:

[0040] The dataset is divided into training, validation, and test sets to construct a crop yield prediction model. The model includes an input layer, a GRU layer, a fully connected layer, and an output layer. The input layer takes historical yield data and the crop growth status over a future n-period as input. A GRU layer processes the historical yield data and the crop growth status over the next n-period, adjusting the number of GRU layers and neurons according to task complexity. A fully connected layer provides additional nonlinear transformations. The output layer outputs the crop yield, using a single neuron to output the predicted value, with an identity function as the activation function. The crop yield prediction model is a gated recurrent unit (GRU) model.

[0041] The model uses mean absolute error as the loss function to measure the error between the model's predicted and actual values; it trains the model using training data and minimizes the loss function using the Adam optimizer; it evaluates the model's performance using a validation set and tunes the model's hyperparameters until the model's performance no longer improves or reaches a preset stopping condition; and it evaluates the model's performance in the prediction task using a test set by inputting the current yield data and the crop's growth status over the next n periods into the pre-trained crop yield prediction model to obtain the crop yield.

[0042] Preferably, the method for assessing whether a crop is in its optimal growth state based on the predicted crop yield includes:

[0043] A preset crop yield threshold is set, and the predicted crop yield is compared with the preset crop yield threshold.

[0044] If the predicted crop yield is less than the preset crop yield threshold, the crop is determined to be in the best growth state.

[0045] If the predicted crop yield is greater than or equal to the preset crop yield threshold, the crop is determined not to be in its optimal growth state.

[0046] The technical effects and advantages of the crop growth status analysis and yield evaluation method of this invention are as follows:

[0047] This invention, through weighted moving averages and sliding window adjustments, effectively removes short-term fluctuations caused by environmental changes and observation errors, resulting in more stable crop growth data that facilitates subsequent analysis. Smoothing helps suppress random fluctuations caused by accidental factors (such as short-term climate changes and instantaneous environmental fluctuations), allowing the data to better reflect long-term trends and patterns. The sliding window size dynamically adjusts according to changes in the growth environment data, enabling flexible responses to different environmental fluctuations. The adaptive window adjustment mechanism ensures the smoothing process is closely integrated with actual environmental changes, improving the filter's adaptability and model stability. By combining environmental weights and difference adjustment coefficients, crop growth data can be weighted and adjusted according to current environmental conditions, further enhancing the rationality and accuracy of data smoothing. Smoothing reduces unnecessary interference, allowing the model to more accurately reflect the true growth trend of crops. This is crucial for subsequent tasks such as growth prediction and pest and disease early warning. Accurate crop growth data provides strong support for agricultural decision-making, helping to formulate more scientific planting strategies, climate response measures, and other optimization decisions.

[0048] By combining multiple environmental factors such as meteorological data, soil data, air quality, and pests and diseases, it is possible to comprehensively assess crop growth under different environmental conditions and dynamically adjust the smoothing strategy according to changes in these factors, thereby enhancing the system's adaptability. By controlling the constants that affect the sliding window size due to environmental fluctuations and the adaptive mechanism of the adjustment coefficient, the sensitivity to different types of environmental changes can be finely adjusted to better cope with complex growth environments. The smoothed data can be used for more stable and reliable analysis, avoiding misleading results caused by data fluctuations, thus enhancing the model's predictive performance and reliability.

[0049] Kalman filtering effectively reduces noise introduced by factors such as measurement errors and environmental fluctuations, generating more accurate and reliable crop growth data. This makes crop growth trends and patterns more apparent, facilitating subsequent analysis. Dynamically adjusting the Kalman gain, through feedback and covariance matrix updates, can optimize the balance between prediction and actual observation in real time, making crop growth predictions more closely reflect reality. As growth environment data changes, the Kalman gain can be dynamically adjusted to optimize the prediction model's response to environmental changes. During crop growth monitoring, adaptive adjustment of the Kalman gain helps the system better cope with sudden environmental changes (such as climate change and soil moisture), providing real-time and accurate crop growth assessments. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating a method for analyzing crop growth status and evaluating yield.

[0051] Figure 2 This is a schematic diagram of a crop growth status analysis and yield evaluation system. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Example 1

[0054] Please see Figure 1 As shown in this embodiment, a method for analyzing crop growth status and evaluating yield includes:

[0055] S1. Real-time capture of crop growth monitoring data through IoT edge node clusters; crop growth monitoring data includes crop growth data, growth environment data, and crop growth image data;

[0056] S2. An environmental weighting mechanism is introduced. The environmental weight of each time point is calculated based on the time series growth environment data, and the adjustment coefficient of the weight influence is dynamically adjusted through the deviation feedback mechanism. The improved moving average algorithm is used in combination with the growth environment data to smooth the crop growth data. The window size is automatically adjusted according to environmental fluctuations to obtain smoothed crop growth data.

[0057] The Kalman filter algorithm is used to filter out noise from the smoothed crop growth data and refine it into crop growth analytical data; the growth environment data and crop growth image data are processed to obtain growth environment variable data and growth image analytical data.

[0058] S3. The crop growth analysis data, growth environment variable data and growth image analysis data are fused to obtain the crop growth status aggregate;

[0059] S4. Input the crop growth status aggregate into the pre-trained growth status assessment model to predict the crop growth status over the next n periods.

[0060] S5. Obtain historical yield data, train a crop yield prediction model based on historical yield data and the crop growth trend over the next n periods, and predict the crop yield based on the crop yield prediction model.

[0061] S6. Based on the predicted crop yield, assess whether the crop is in its optimal growth condition.

[0062] S7. If crops are not in their optimal growth state, feedback will be sent to staff through the intelligent crop growth monitoring terminal for timely intervention.

[0063] Crop growth data includes growth morphology data and growth function data; growth environment data includes meteorological data, soil data, air quality data, and pest and disease data; crop growth image data includes crop vegetation index images, chlorophyll content images, plant height images, crop canopy cover images, and leaf area index images.

[0064] Growth morphology data includes crop height, stem diameter, and root depth; stem diameter and root depth include leaf area index, photosynthetic efficiency, and canopy coverage; meteorological data includes growth temperature, precipitation, humidity, light intensity, and sunshine duration; soil data includes soil moisture content, soil temperature, soil nutrient content, soil pH, and soil salinity; air quality data includes atmospheric carbon dioxide concentration and pollutant concentration; pest and disease data includes disease data and pest data; disease data includes disease types, disease distribution areas, and crop incidence rates; pest data includes pest types, pest density, and pest distribution areas; soil nutrient content includes the concentrations of elements such as nitrogen, phosphorus, and potassium; the IoT edge node cluster includes distributed soil sensors, leaf microenvironment probes, UAV-borne hyperspectral imagers, and satellite remote sensing terminals.

[0065] Methods for obtaining smoothed crop growth data include:

[0066] S31. Preset growth morphology data are: in, This refers to the growth morphology data at time point t; x q Let q be the morphological index (such as crop height, leaf area index, root depth, etc.) in the growth morphology data at time point t; q is the index of the morphological index, q = 1,...,N; N is the total number of morphological indexes.

[0067] The preset growth function data is as follows: in, For growth function data at time point t; y j Let j be the j-th functional index in the growth functional data at time point t; j is the index of the functional index, j = 1, ..., M; M is the total number of functional indices;

[0068] S32, Growth morphology data at time point t Growth function data at time point t The data is pieced together to form crop growth data, which is represented as: x t ={x1,x2,...,x q ,...,x N ,y1,y2,...,y j ,...,y M}; where x t This refers to crop growth data at time point t;

[0069] The preset growth environment data at time point t are: in, For meteorological data at time point t; Soil data at time point t; This refers to air quality data at time point t; This refers to pest and disease data at time point t;

[0070] S33. Based on the growth environment data E at time point t t The environmental weight at each time point t is calculated using the environmental weight calculation formula; the environmental weight calculation formula is: Where, ω i (E t E represents the growth environment data at time point t. t The weight corresponding to the i-th time step determines the importance of environmental factors at that moment in the weighted moving average. The larger the weight, the greater the impact of the environmental factors at that time on the final calculation result; E max For ideal growth environment data, this includes growth temperature, humidity, light intensity, soil moisture content, etc. For example, at time point t, the ideal growth environment data is 25℃ and humidity is 80%, then E max ={25, 80}; ||E t -E max || represents the growth environment data and ideal growth environment data E corresponding to time point t. max The difference between the current and ideal environmental conditions measures the deviation from the current conditions; the greater the deviation, the greater the influence of the factors on the weight. ΔE t This represents the rate of change of the growth environment data from time point t-1 to time point t; α is the ratio of the control growth environment data to the ideal growth environment data E. max The coefficient for adjusting the impact of the difference between the two on the weights; β is the coefficient for adjusting the impact of the rate of change of the growth environment data on the weights; i is the index of the time step;

[0071] It should be noted that the design basis of the environmental weight calculation formula includes: considering the difference between the environment and the ideal state: crop growth is significantly affected by the environment, the ideal growth environment is conducive to crop growth, but the actual environment differs from this. The formula introduces ||E t -E max ||, by calculating actual growth environment data E t Data E of the ideal growth environment max The difference in environmental weights measures the degree to which the current environment deviates from an ideal state. For example, ideal temperatures are conducive to photosynthesis and nutrient absorption; the greater the deviation of the actual temperature from this ideal, the greater the potential negative impact on crop growth. This difference should be reflected when calculating environmental weights.

[0072] Pay attention to the rate of environmental change: Environmental change has a significant impact on crop growth; a stable environment is conducive to growth, while a highly volatile environment may cause stress responses. ΔEt This represents the rate of change of the growing environment data from time t-1 to time t, reflecting environmental changes. Sudden temperature changes or large fluctuations in precipitation can affect crop physiological processes. Incorporating this parameter into the environmental weight calculation formula allows for timely adjustment of weights based on environmental changes.

[0073] The introduction of adjustment coefficients: Different crops have different sensitivities to environmental differences and changes. By adjusting the coefficients α and β, we can adapt to the growth characteristics and actual needs of different crops.

[0074] By comprehensively considering the differences between the current environment and the ideal state, as well as the rate of environmental change, environmental weights are accurately calculated to reflect the suitability and potential impact of the current environment on crop growth. Higher weights indicate a more ideal environment or more stable changes, which is beneficial to crop growth; conversely, lower weights are detrimental. When smoothing crop growth data, incorporating environmental weights allows for the assignment of different weights to data at different time points based on environmental conditions. When the environment is suitable and stable, the corresponding data has a higher weight and a greater impact on the smoothing results; when the environment is unfavorable or highly variable, the data has a lower weight, reducing its interference with the smoothing results and allowing the smoothed data to better reflect the true growth trend of crops. The environmental weight calculation formula comprehensively considers the differences between the current environment and the ideal state, as well as the rate of environmental change, to fully reflect the environmental impact on crop growth. Compared to measuring environmental impact by a single factor, this method is more consistent with reality and provides a more accurate basis for crop growth analysis.

[0075] Compared to existing technologies, the technical effects are as follows:

[0076] More accurate environmental impact assessment: Existing technologies may not fully consider the differences between the environment and the ideal state, as well as the rate of environmental change, or may only consider a single factor. The environmental weight calculation formula comprehensively integrates these factors, enabling a more accurate assessment of the environmental impact on crop growth and providing more accurate information for agricultural decision-making. For example, when determining whether measures need to be taken to improve the environment, assessments based on this formula are more valuable. Optimized data processing: Combining environmental weights with data smoothing can more effectively remove noise and interference, highlighting the true trend of crop growth. When predicting crop growth status and yield, models trained on data processed by this formula are more accurate, improving the reliability of predictions. For example, when predicting crop yield, it can more accurately reflect the impact of environmental factors on yield, reducing prediction errors. S34. Adaptive formula for difference adjustment coefficients is used to compare the controlled growth environment data with the ideal growth environment data E. max The adjustment coefficient α, which modifies the degree of influence of the difference on the weight, is dynamically adjusted. The adaptive formula for the difference adjustment coefficient is: Where m represents the number of types of growth environment data; denoted as , where is the rate of change of the c-th growth environment data at time t; b is a constant adjusting the influence of the rate of change on the adjustment coefficient; p is an exponential factor controlling the cumulative effect of the rate of change on the adjustment coefficient; and c is the index of the growth environment data type, ranging from 1 to m.

[0077] The adjustment coefficient α is dynamically adjusted based on the rate of change of environmental data, ensuring that it automatically decreases in rapidly changing environments, thus avoiding over-reliance on environmental changes. Conversely, when the environment is stable, the adjustment coefficient can be maintained at a higher level, enhancing its influence on the weights. Through the constant b and the exponent p, the adaptive formula for the differential adjustment coefficient can be flexibly adjusted according to actual environmental needs, adapting to different types and rates of change in growth environment data. It exhibits good adaptability in different agricultural scenarios, effectively incorporating the types and changes of environmental factors, and achieving reasonable smoothing of crop growth data through the adjustment coefficient.

[0078] Different degrees of environmental change require different adjustment coefficients α to balance the impact of environmental differences on the weighting of crop growth data. When environmental changes are small, it is desirable to maintain a high level of α to maximize the influence of environmental differences on the weighting; when environmental changes are drastic, α should be reduced to avoid over-reliance on environmental change data. The formula uses two parameters, b and p, to flexibly adjust α. By dynamically adjusting α, the calculation of environmental weights is improved, the weighted moving average process is optimized, and the quality of crop growth data smoothing is enhanced. This allows the smoothed data to more accurately reflect the true growth trend of crops, providing reliable data support for subsequent growth status analysis and yield prediction.

[0079] Compared to existing technologies, the technical effects are as follows:

[0080] Enhancing Environmental Adaptability: Existing technologies may employ fixed adjustment coefficients or simple adjustment methods, which cannot adapt to complex and ever-changing environments. This formula can dynamically adjust α in real time according to environmental changes, making the system's response to environmental changes more sensitive and accurate, and enhancing the adaptability of the crop growth status analysis and yield evaluation system to different environmental conditions.

[0081] Improving data processing accuracy: In terms of data processing, this formula optimizes the calculation of environmental weights, making the weighted moving average more accurate, effectively reducing the interference of environmental fluctuations on the data, and improving the data smoothing effect. Based on the processed data, growth status prediction and yield assessment can be performed, significantly improving the accuracy of prediction and assessment.

[0082] For example, the initial adjustment coefficient α is 0.8, the number of growth environment data types m is 3, and the rate of change for each growth environment data is respectively... The constant b, which adjusts the influence of the rate of change on the adjustment coefficient, is 2; the exponential factor p, which controls the cumulative effect of the rate of change on the adjustment coefficient, is 1; the cumulative sum of environmental change rates is calculated as follows: So after dynamic adjustment

[0083] S35. Use the weighted moving average formula to smooth the crop growth data to obtain smoothed crop growth data; the weighted moving average formula is: Among them, WA t The data represents smoothed crop growth data at time point t; x t-a γ represents the crop growth data at time point ta, which is the unsmoothed growth data used to generate the smoothed result; γ is the sliding window size; a is the index of the sliding window, with a value ranging from 0 to γ-1.

[0084] S36. Dynamically adjust the size γ of the sliding window using a sliding window adjustment formula. The sliding window adjustment formula is: Where, γ max ω represents the maximum value of the sliding window. avg is the average weight of the growth environment data; d is a constant that controls the effect of environmental fluctuations on the sliding window size; This is the floor function;

[0085] The sliding window adjustment formula dynamically adjusts the size of the sliding window according to changes in the environment in order to better cope with external changes. When the environment fluctuates greatly, the window becomes smaller, which helps to capture subtle changes; while when the environment is relatively stable, the window is larger, which helps to smooth out noise in the data. By controlling the adjustment of the constant d that controls the impact of environmental fluctuations on the sliding window size, the window is reduced when the environment fluctuates greatly to adapt to rapidly changing data, and the window is increased when the environment is relatively stable to ensure data smoothness.

[0086] By dynamically adjusting the sliding window size, the data processing can better adapt to different environmental fluctuations. When the environment changes drastically, the window size is reduced to enhance the ability to capture data details; when the environment is stable, the window size is increased to improve the smoothing effect of the data, thereby more accurately reflecting the characteristics of crop growth data and providing more reliable data support for subsequent analysis and prediction.

[0087] Optimizing the effect of weighted moving averages: Properly adjusting the sliding window size helps optimize the effect of weighted moving averages. An appropriate window size allows the weighted moving average to remove noise while preserving the most useful information in the data, avoiding data distortion or misjudgment of trends caused by inappropriate window size, and improving the quality of crop growth data smoothing.

[0088] Compared to existing technologies, the technical effects are as follows:

[0089] Enhancing Data Processing Flexibility: Existing technologies typically use a fixed sliding window size, which cannot adaptively adjust to changes in the environment. The sliding window adjustment formula of this invention can dynamically change the window size, enabling more effective processing of crop growth data under different environmental conditions, thus improving the flexibility and adaptability of data processing.

[0090] Improving forecast accuracy: Dynamically adjusted sliding windows can better process data, making subsequent crop growth forecasts and yield assessments based on this data more accurate. By more accurately capturing data changes and trends, forecast errors caused by improper data processing are reduced, providing a more reliable basis for agricultural production decisions and helping to improve the efficiency and effectiveness of agricultural production.

[0091] Adapting to Complex Environmental Changes: When faced with complex and ever-changing crop growth environments, this formula can respond promptly to environmental changes and adjust the window size. In contrast, the fixed window of existing technologies is difficult to adapt to such complex environments and is prone to data processing errors. The formula of this invention can better adapt to complex environmental changes, ensuring the stability and reliability of the crop growth analysis and yield evaluation system.

[0092] For example, the maximum value γ of the sliding window. max The value is 10, and the growth environment data E at time point t is... t The weight ω corresponding to the i-th time step i (E t The weight of the growth environment data is 7; the average weight of the growth environment data is 5; the constant d for controlling the effect of environmental fluctuations on the sliding window size is 0.2, then the sliding window size is...

[0093] Methods for using the Kalman filter algorithm to filter out noise from smoothed crop growth data and refine it into analytical crop growth data include:

[0094] Define a crop growth state transition model to reflect the change process of crop growth state over time. Based on the crop growth state at the previous time point and the control input, predict the crop growth state at the current time point.

[0095] The crop growth state transition model is as follows: in, is the predicted value of the crop growth state at the current time point t; A is the state transition matrix, which represents the relationship between the crop growth state from time point t-1 to time point t. is the predicted value of crop growth status at time point t-1; B is the control input matrix, representing the external control input vector u. t Impact on crop growth status; u t For external control input vector;

[0096] Crop growth exhibits specific patterns and biological characteristics, with inherent connections between changes in various indicators at different growth stages. Taking corn growth as an example, the changes in growth indicators such as plant height, number of leaves, and stem diameter follow certain patterns at different growth stages. Researchers, through long-term observation and research on the corn growth process, have grasped the relationships between these indicators at adjacent time points. For instance, during the corn jointing stage, the current plant height is correlated with the previous plant height and number of leaves, because photosynthesis in the leaves provides energy and substances for plant height growth. By mathematically modeling this inherent relationship, the influence relationships between various growth indicators are represented in the form of matrix elements, thus constructing a state transition matrix A. In agricultural production, external control measures such as irrigation, fertilization, and pest and disease control have a significant impact on crop growth. Agricultural experts, through long-term practical experience, understand the appropriate dosage and timing of different control measures at different crop growth stages, as well as the degree of impact of these measures on crop growth status. For example, during the tillering stage of rice growth, appropriate nitrogen fertilizer application can promote tillering and increase the number of effective panicles. Based on this experience, external control inputs (such as fertilizer application and irrigation) are correlated with crop growth status indicators, and the influence coefficients of different control inputs on each growth status indicator are determined. These coefficients constitute the elements of the control input matrix B.

[0097] The predicted crop growth status at the current time point t is updated using a state update formula, which is: in, K represents the predicted value of the updated crop growth status. t The Kalman gain controls the balance between the predicted and actual observed crop growth states; z t H represents the actual observed values ​​of crop growth status; H is the observation matrix, which represents the mapping from the predicted values ​​of crop growth status to the actual observed values ​​of crop growth status.

[0098] The Kalman gain is calculated using the Kalman gain formula, which is used to balance the weights between the predicted and actual observed values ​​of crop growth status. The Kalman gain formula is as follows: Among them, P t R is the prediction error covariance matrix; t (ΔE t H represents the measurement noise covariance matrix that is dynamically adjusted based on the rate of change of growth environment data. TThis is the transpose of the observation matrix;

[0099] The error covariance matrix is ​​updated using a covariance matrix update formula to reflect the uncertainty of the updated crop growth status; the covariance matrix update formula is: in, Let I be the updated error covariance matrix; I is the identity matrix; the identity matrix I is a special square matrix. Its main diagonal elements are all 1s, and all other elements are 0s.

[0100] The crop growth data after denoising through Kalman filtering is used as the final output to obtain crop growth analysis data.

[0101] Methods for processing growth environment data and crop growth image data to obtain growth environment variable data and growth image analysis data include:

[0102] Density clustering algorithm is used to identify and remove outliers in crop growth monitoring data, including growth environment data and crop growth image data, to obtain growth environment variable data and growth image analysis data.

[0103] Methods for obtaining crop growth state aggregates include:

[0104] Normalized crop growth analysis data, growth environment variable data, and growth image analysis data are fused using a weighted model to obtain a composite of crop growth status.

[0105] Crop growth analysis data is denoted as O1, growth environment variable data as O2, and growth image analysis data as O3. The weighted model is: QZ = O1·δ1 + O2·δ2 + O3·δ3; where QZ is the aggregate of crop growth status; δ1 is the weight coefficient of crop growth analysis data; δ2 is the weight coefficient of growth environment variable data; and δ3 is the weight coefficient of growth image analysis data.

[0106] Based on the importance of the normalized crop growth analysis data in the crop growth status aggregate, the initial value range of the weighting factor δ1 is set to 0.2–0.4, with an initial default value of 0.2. Using historical crop growth status aggregates, the influence of the normalized crop growth analysis data on the crop growth status is analyzed. If the normalized crop growth analysis data has a significant impact on the crop growth status (e.g., the proportion affecting the evaluation results exceeds 50%), the value of δ1 is increased; otherwise, its value is decreased.

[0107] Based on the importance of normalized growth environment variables in the crop growth status aggregate, the initial value range of the weighting factor δ2 is set to 0.4–0.6, with an initial default value of 0.5. Using historical crop growth status aggregates, the influence of normalized growth environment variables on crop growth status is analyzed. If the normalized growth environment variables have a significant impact on crop growth status (e.g., changes in growth environment characteristics directly affect crop growth status), the value of δ2 is increased; otherwise, its value is decreased.

[0108] Based on the importance of the normalized growth image analysis data in the crop growth status aggregate, the initial value range of the weighting factor δ3 is set to 0.3–0.5, with an initial default value of 0.3. The degree of influence is determined by analyzing the correlation between the normalized growth image analysis data and the crop growth status (e.g., the role of growth image analysis data in identifying crop growth status). If the normalized growth image analysis data plays a key role in the actual crop growth process, the value of δ3 is increased; if its influence is weak, it is appropriately decreased.

[0109] Based on real-time feedback from the model, the values ​​of weight coefficients δ1, δ2, and δ3 are gradually optimized to improve the model's generalization ability.

[0110] Training methods for growth status assessment models include:

[0111] The dataset is divided into training, validation, and test sets for training and evaluating the model. A sample set is a subset of the dataset, each containing an aggregate of historical crop growth states and the corresponding crop growth status over the next n time periods. A growth status evaluation model is constructed using the TensorFlow deep learning library. The growth status evaluation model includes an input layer, an LSTM layer, and an output layer. The input layer takes the aggregate of historical crop growth states as input, and the output layer outputs the crop growth status over the next n time periods. The growth status evaluation model is an LSTM model.

[0112] Define the model's loss function, using the L2 regularized mean squared error loss function to measure the difference between the model's predicted values ​​and the true values; train the growth status assessment model using the training set, update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the growth status assessment model by calculating the accuracy metric.

[0113] The Adam optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback of the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The pre-trained growth status evaluation model was used to predict the current crop growth status aggregate to obtain the crop growth status in the next n time periods.

[0114] Historical yield data includes historical crop yield per unit area, historical total crop yield, historical crop yield under different environments, and historical yield fluctuation data. Historical yield data is obtained by accessing the historical crop database through the intelligent crop growth monitoring terminal, which interacts with the historical crop database through a standardized API interface.

[0115] The methods for constructing a historical crop database include: collecting crop yield per unit area in real time through sensors installed in farmland, combining planting and harvesting information manually recorded by farmers and annual agricultural production reports from government statistical departments, and using mature database management systems (such as MySQL and PostgreSQL) to construct the historical crop database.

[0116] The historical crop database is used to store historical crop yield per unit area, historical total crop yield, historical crop yield under different environments, and historical yield fluctuation data. Through API interfaces and IoT gateways, new data is synchronized to the database in real time, and historical data query and retrieval functions are configured. Users can directly access the database information through intelligent crop growth monitoring terminals.

[0117] Training methods for crop yield prediction models include:

[0118] The dataset is divided into training, validation, and test sets to construct a crop yield prediction model. The sample set is a subset of the dataset, and each sample set includes historical yield data and the crop growth status and corresponding crop yield over the next n periods. The crop yield prediction model includes an input layer, a GRU layer, a fully connected layer, and an output layer.

[0119] The input layer of the crop yield prediction model is used to input historical yield data and the crop growth status over the next n periods. A GRU layer is used to process the historical yield data and the crop growth status over the next n periods. The number of GRU layers and neurons is adjusted according to the task complexity, and a fully connected layer is used to provide additional nonlinear transformations. The model output layer outputs the crop yield, using a single neuron to output the predicted value, and an identity function as the activation function. The crop yield prediction model is a gated recurrent unit model.

[0120] The model uses mean absolute error as the loss function to measure the error between the model's predicted and actual values; it trains the model using training data and minimizes the loss function using the Adam optimizer; it evaluates the model's performance using a validation set and tunes the model's hyperparameters until the model's performance no longer improves or reaches a preset stopping condition; and it evaluates the model's performance in the prediction task using a test set by inputting the current yield data and the crop's growth status over the next n periods into the pre-trained crop yield prediction model to obtain the crop yield.

[0121] Methods for assessing whether crops are in optimal growth condition based on predicted crop yields include:

[0122] A preset crop yield threshold is set, and the predicted crop yield is compared with the preset crop yield threshold.

[0123] If the predicted crop yield is less than the preset crop yield threshold, the crop is determined to be in the best growth state.

[0124] If the predicted crop yield is greater than or equal to the preset crop yield threshold, the crop is determined not to be in its optimal growth state.

[0125] The preset crop yield threshold is set by staff. The yield of different crops is collected through a smart crop growth monitoring terminal, and the average yield of multiple crops is taken as the preset crop yield threshold.

[0126] This embodiment effectively removes short-term fluctuations caused by environmental changes and observation errors through weighted moving averages and sliding window adjustments, resulting in more stable crop growth data that facilitates subsequent analysis. Smoothing helps suppress random fluctuations caused by accidental factors (such as short-term climate changes and instantaneous environmental fluctuations), allowing the data to better reflect long-term trends and patterns. The sliding window size dynamically adjusts according to changes in the growth environment data, enabling flexible responses to different environmental fluctuations. The adaptive window adjustment mechanism ensures the smoothing process is closely integrated with actual environmental changes, improving the filter's adaptability and model stability. By combining environmental weights and difference adjustment coefficients, crop growth data can be weighted and adjusted according to current environmental conditions, further enhancing the rationality and accuracy of data smoothing. Smoothing reduces unnecessary interference, allowing the model to more accurately reflect the true growth trend of crops. This is crucial for subsequent tasks such as growth prediction and pest and disease early warning. Accurate crop growth data provides strong support for agricultural decision-making, helping to formulate more scientific planting strategies, climate response measures, and other optimization decisions.

[0127] By combining multiple environmental factors such as meteorological data, soil data, air quality, and pests and diseases, it is possible to comprehensively assess crop growth under different environmental conditions and dynamically adjust the smoothing strategy according to changes in these factors, thereby enhancing the system's adaptability. By controlling the constants that affect the sliding window size due to environmental fluctuations and the adaptive mechanism of the adjustment coefficient, the sensitivity to different types of environmental changes can be finely adjusted to better cope with complex growth environments. The smoothed data can be used for more stable and reliable analysis, avoiding misleading results caused by data fluctuations, thus enhancing the model's predictive performance and reliability.

[0128] Kalman filtering effectively reduces noise introduced by factors such as measurement errors and environmental fluctuations, generating more accurate and reliable crop growth data. This makes crop growth trends and patterns more apparent, facilitating subsequent analysis. Dynamically adjusting the Kalman gain, through feedback and covariance matrix updates, can optimize the balance between prediction and actual observation in real time, making crop growth predictions more closely reflect reality. As growth environment data changes, the Kalman gain can be dynamically adjusted to optimize the prediction model's response to environmental changes. During crop growth monitoring, adaptive adjustment of the Kalman gain helps the system better cope with sudden environmental changes (such as climate change and soil moisture), providing real-time and accurate crop growth assessments.

[0129] Example 2

[0130] Please see Figure 2 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. A crop growth status analysis and yield evaluation system is provided, including:

[0131] The data acquisition unit is used to capture crop growth monitoring data in real time through an IoT edge node cluster; the crop growth monitoring data includes crop growth data, growth environment data, and crop growth image data.

[0132] The data processing unit is used to introduce an environmental weighting mechanism, calculate the environmental weight at each time point based on the time series growth environment data, and dynamically adjust the adjustment coefficient of the weight influence through a deviation feedback mechanism; it uses an improved moving average algorithm combined with growth environment data to smooth the crop growth data, and automatically adjusts the window size according to environmental fluctuations to obtain smoothed crop growth data.

[0133] The Kalman filter algorithm is used to filter out noise from the smoothed crop growth data and refine it into crop growth analytical data; the growth environment data and crop growth image data are processed to obtain growth environment variable data and growth image analytical data.

[0134] The data fusion unit is used to fuse crop growth analysis data, growth environment variable data, and growth image analysis data to obtain a composite of crop growth status.

[0135] The growth status analysis unit is used to import the crop growth status aggregate into the pre-trained growth status assessment model to predict the crop growth status over the next n periods.

[0136] The yield prediction unit is used to acquire historical yield data, train a crop yield prediction model based on the historical yield data and the crop growth status over the next n periods, and predict the crop yield based on the crop yield prediction model.

[0137] The growth status assessment unit is used to assess whether crops are in optimal growth status based on predicted crop yields.

[0138] The feedback unit is optimized so that if crops are not in their optimal growth state, feedback is sent to staff through the intelligent crop growth monitoring terminal for timely intervention.

[0139] Since the electronic device described in this embodiment is used to implement the crop growth status analysis and yield evaluation method described in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the crop growth status analysis and yield evaluation method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any electronic device used by those skilled in the art to implement the crop growth status analysis and yield evaluation method described in this application embodiment falls within the scope of protection of this application.

[0140] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0141] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for analyzing crop growth status and evaluating yield, characterized in that, include: S1. Real-time capture of crop growth monitoring data through IoT edge node clusters; crop growth monitoring data includes crop growth data, growth environment data, and crop growth image data; S2. An environmental weighting mechanism is introduced. The environmental weight of each time point is calculated based on the time series growth environment data, and the adjustment coefficient of the weight influence is dynamically adjusted through the deviation feedback mechanism. The improved moving average algorithm is used in combination with the growth environment data to smooth the crop growth data. The window size is automatically adjusted according to environmental fluctuations to obtain smoothed crop growth data. The Kalman filter algorithm is used to filter out noise from the smoothed crop growth data and refine it into crop growth analytical data; the growth environment data and crop growth image data are processed to obtain growth environment variable data and growth image analytical data. The method for obtaining the smoothed crop growth data includes: S31. Preset growth morphology data are: ;in, For at a certain point in time Growth morphology data; For at a certain point in time The first in the growth morphology data Individual morphological indicators; For indexing morphological indicators, ; This represents the total number of morphological indicators; The preset growth function data is as follows: ;in, For at a certain point in time Growth function data; For at a certain point in time The first in the growth function data One functional indicator; An index for functional metrics. ; The total number of functional indicators; S32, at the time point Growth morphology data At a certain point in time Growth function data By combining these data, we can form crop growth data, which is represented as follows: ;in, For at a certain point in time Crop growth data; Preset at time point The growth environment data are as follows: ;in, For at a certain point in time Meteorological data; For at a certain point in time Soil data; For at a certain point in time Air quality data; For at a certain point in time Data on pests and diseases; S33, Based on time points Growth environment data The environmental weight calculation formula is used to calculate the weight of each time point. Environmental weight; the formula for calculating environmental weight is: ;in, For at a certain point in time Growth environment data Corresponding to the The weight of each time step; Data for an ideal growth environment; For time points Corresponding growth environment data and ideal growth environment data The differences between them; Indicates from a point in time At the appointed time The rate of change of growth environment data; To control growth environment data and ideal growth environment data The adjustment coefficient for the degree of influence of the difference between them on the weight; An adjustment coefficient to control the impact of the rate of change in growth environment data on the weighting; For the index of the time step; S34. Using an adaptive formula with a difference adjustment coefficient, the control growth environment data and the ideal growth environment data are compared. The adjustment coefficient for the degree of influence of the difference between them on the weight. Dynamic adjustments are made, and the adaptive formula for the difference adjustment coefficient is as follows: ;in, The number of types of growth environment data; For the first Growth environment data at time points The rate of change; A constant used to adjust the influence of the rate of change on the adjustment coefficient; An exponential factor used to control the impact of the cumulative rate of change on the adjustment coefficient; An index for the types of growth environment data; S35. Use the weighted moving average formula to smooth crop growth data; the weighted moving average formula is: ;in, For time points Smoothed crop growth data; For time points Crop growth data, For time points Environmental weights; To adjust the sliding window size; The index of the sliding window; S36. Dynamically adjust the size of the sliding window using the sliding window adjustment formula. The formula for adjusting the sliding window is: ;in, This represents the maximum value of the sliding window. The average weight of the growth environment data; A constant used to control the effect of environmental fluctuations on the sliding window size; This is the floor function; S3. The crop growth analysis data, growth environment variable data and growth image analysis data are fused to obtain the crop growth status aggregate; S4. Input the crop growth status aggregate into the pre-trained growth status assessment model to predict the crop growth status over the next n periods. S5. Obtain historical yield data, train a crop yield prediction model based on historical yield data and the crop growth trend over the next n periods, and predict the crop yield based on the crop yield prediction model. S6. Based on the predicted crop yield, assess whether the crop is in its optimal growth condition. S7. If crops are not in their optimal growth state, feedback will be sent to staff through the intelligent crop growth monitoring terminal for timely intervention.

2. The method for analyzing crop growth status and evaluating yield according to claim 1, characterized in that, The crop growth data includes growth morphology data and growth function data; the growth environment data includes meteorological data, soil data, air quality data, and pest and disease data; the crop growth image data includes crop vegetation index images, chlorophyll content images, plant height images, crop canopy coverage images, and leaf area index images.

3. The method for analyzing crop growth status and evaluating yield according to claim 2, characterized in that, The method for using the Kalman filter algorithm to filter out noise from smoothed crop growth data and refine it into analytical crop growth data includes: Define a crop growth state transition model to reflect the change process of crop growth state over time. Based on the crop growth state at the previous time point and the control input, predict the crop growth state at the current time point. By updating the current time point The predicted values ​​of crop growth status are obtained, and the Kalman gain is calculated to balance the weights between the predicted and actual observed values ​​of crop growth status. The error covariance matrix in the Kalman gain is updated to reflect the uncertainty of the updated crop growth status. The crop growth data after denoising by Kalman filtering is output as the final result to obtain the crop growth analysis data.

4. The method for analyzing crop growth status and evaluating yield according to claim 3, characterized in that, The method for processing growth environment data and crop growth image data to obtain growth environment variable data and growth image analysis data includes: Density clustering algorithm is used to identify and remove outliers in crop growth monitoring data, including growth environment data and crop growth image data, to obtain growth environment variable data and growth image analysis data.

5. The method for analyzing crop growth status and evaluating yield according to claim 4, characterized in that, The method for obtaining the crop growth state aggregate includes: The crop growth analysis data, growth environment variable data, and growth image analysis data were normalized by standard deviation, and converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, to obtain normalized crop growth analysis data, growth environment variable data, and growth image analysis data. The normalized crop growth analysis data, growth environment variable data, and growth image analysis data were then weighted and fused to obtain a crop growth status aggregate.

6. The method for analyzing crop growth status and evaluating yield according to claim 5, characterized in that, The training method for the growth status assessment model includes: The dataset is divided into training, validation, and test sets for training and evaluating the model's performance. A growth status assessment model is constructed using the TensorFlow deep learning library. The growth status assessment model includes an input layer, an LSTM layer, and an output layer. The input layer of the model is used to input the aggregate of historical crop growth states. The output layer of the model is used to output the crop growth status over the next n time periods. The growth status assessment model is an LSTM model. Define the model's loss function, using the L2 regularized mean squared error loss function to measure the difference between the model's predicted values ​​and the true values; train the growth status assessment model using the training set, update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the growth status assessment model by calculating the accuracy metric. The Adam optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback of the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The pre-trained growth status evaluation model was used to predict the current crop growth status aggregate to obtain the crop growth status in the next n time periods.

7. The method for analyzing crop growth status and evaluating yield according to claim 6, characterized in that, The historical production data includes historical crop yield per unit area, historical total crop yield, historical crop yield under different environments, and historical production fluctuation data.

8. The method for analyzing crop growth status and evaluating yield according to claim 7, characterized in that, The training methods for the crop yield prediction model include: The dataset is divided into training, validation, and test sets to construct a crop yield prediction model. The model includes an input layer, a GRU layer, a fully connected layer, and an output layer. The input layer takes historical yield data and the crop growth status over a future n-period as input. A GRU layer processes the historical yield data and the crop growth status over the next n-period, adjusting the number of GRU layers and neurons according to task complexity. A fully connected layer provides additional nonlinear transformations. The output layer outputs the crop yield, using a single neuron to output the predicted value, with an identity function as the activation function. The crop yield prediction model is a gated recurrent unit (GRU) model. The model uses mean absolute error as the loss function to measure the error between the model's predicted value and the actual value; it trains the model using training data and minimizes the loss function using the Adam optimizer; it evaluates the model's performance using a validation set and tunes the model's hyperparameters until the model's performance no longer improves or reaches a preset stopping condition; and it evaluates the model's performance in the prediction task using a test set by inputting the current yield data and the crop's growth status over the next n periods into the pre-trained crop yield prediction model to obtain the crop yield.

9. The method for analyzing crop growth status and evaluating yield according to claim 8, characterized in that, The method for assessing whether crops are in optimal growth condition based on predicted crop yields includes: A preset crop yield threshold is set, and the predicted crop yield is compared with the preset crop yield threshold. If the predicted crop yield is greater than or equal to the preset crop yield threshold, the crop is determined to be in the best growth state. If the predicted crop yield is less than the preset crop yield threshold, the crop is determined to be not in its optimal growth state.

Images