Wheat wintering period seedling quality grading method and system

Abstract

The invention relates to the technical field of seedling cultivation, and discloses a wheat wintering period seedling quality grading method and system, and the method comprises the steps: analyzing an index screening dynamic threshold value corresponding to a morphological index and a physiological and biochemical index; dynamically screening the morphological indexes and the physiological and biochemical indexes to obtain a key index data set; performing principal component analysis on the key index data set to obtain a plurality of principal components and principal component scores; performing clustering analysis on the wheat overwintering period seedlings based on the quality comprehensive value to obtain an initial quality grade of the wheat overwintering period seedlings; the change trend of the wheat overwintering period seedlings is analyzed based on the initial quality grade and the climate change trend, the initial quality grade is optimized according to the change trend, a target quality grade is obtained, and targeted regulation and control measures for the wheat overwintering period seedlings are generated based on the target quality grade. According to the method, the accuracy of wheat wintering period seedling quality grading can be improved.

Description

Technical Field

[0001] This invention relates to the field of seedling cultivation technology, and in particular to a method and system for classifying the quality grade of wheat seedlings during the overwintering period. Background Technology

[0002] With the development of modern agricultural technology, the quantitative analysis of the morphology, physiology, and biochemical characteristics of wheat seedlings during the wintering period, enabling scientific assessment and dynamic regulation of seedling quality, has become a key link in improving the stability of wheat production and the efficiency of resource utilization. Especially against the backdrop of frequent extreme low temperatures and droughts during the wintering period due to climate change, there is an urgent need for a quantifiable and highly adaptable seedling quality assessment method to provide precise technical support for field operations.

[0003] In existing technologies, the assessment of wheat seedling quality during the overwintering period largely relies on manual observation of morphological indicators or the sole use of physiological and biochemical indicators. Manual observation is heavily influenced by subjective experience, leading to low accuracy in the assessment results. Single-indicator assessment methods ignore the synergistic effect between morphological and physiological / biochemical indicators; for example, judging solely by seedling height cannot distinguish between overly vigorous seedlings and robust seedlings, easily causing misjudgments and thus resulting in low accuracy in classifying the quality of wheat seedlings during the overwintering period. Summary of the Invention

[0004] This invention provides a method and system for classifying the quality grade of wheat seedlings during the overwintering period, the main purpose of which is to solve the problem of low accuracy in classifying the quality grade of wheat seedlings during the overwintering period.

[0005] To achieve the above objectives, the present invention provides a method for classifying the quality grade of wheat seedlings during the overwintering period, comprising: Morphological and physiological-biochemical indicators of wheat seedlings during the overwintering period were collected, and dynamic thresholds for the corresponding indicators were screened using a preset multi-objective optimization model. The morphological indicators and the physiological and biochemical indicators are dynamically filtered according to the dynamic threshold of the indicators to obtain a key indicator dataset; Principal component analysis was performed on the key indicator dataset to obtain multiple principal components and their corresponding principal component scores. The principal component scores are standardized to obtain the comprehensive quality value of wheat seedlings during the wintering period. Based on the comprehensive quality value, cluster analysis is performed on the wheat seedlings during the wintering period to obtain the initial quality grade of the wheat seedlings during the wintering period. Based on the initial quality level and the preset climate change trend analysis, the changing trend of wheat seedlings during the wintering period is analyzed. The initial quality level is optimized according to the changing trend to obtain the target quality level. Based on the target quality level, targeted control measures for wheat seedlings during the wintering period are generated.

[0006] To address the above problems, the present invention also provides a wheat overwintering seedling quality grading system, the system comprising: The indicator screening dynamic threshold analysis module is used to collect morphological and physiological and biochemical indicators of wheat seedlings during the wintering period, and to analyze the indicator screening dynamic thresholds corresponding to the morphological and physiological and biochemical indicators using a preset multi-objective optimization model. The key indicator dataset generation module is used to dynamically filter the morphological indicators and the physiological and biochemical indicators according to the indicator screening dynamic threshold to obtain the key indicator dataset. The principal component analysis module is used to perform principal component analysis on the key indicator dataset to obtain multiple principal components and the principal component scores corresponding to the multiple principal components. The initial quality grade analysis module is used to standardize the principal component scores to obtain the comprehensive quality value of wheat seedlings during the wintering period, and to perform cluster analysis on the wheat seedlings during the wintering period based on the comprehensive quality value to obtain the initial quality grade of the wheat seedlings during the wintering period. The regulation measure generation module is used to analyze the changing trend of wheat seedlings during the wintering period based on the initial quality level and the preset climate change trend, optimize the initial quality level according to the changing trend to obtain the target quality level, and generate targeted regulation measures for wheat seedlings during the wintering period based on the target quality level.

[0007] This invention provides comprehensive data for subsequent analysis by collecting multi-dimensional indicators; the dynamic thresholds determined by the multi-objective optimization model avoid the poor adaptability problem of fixed thresholds, improving the flexibility and accuracy of indicator selection; the combination of principal component loading analysis and forward selection strategy enables precise selection of key indicators; principal component analysis achieves dimensionality reduction of multiple indicators, providing a simple and efficient data foundation for calculating the comprehensive quality value; the comprehensive quality value integrates information from multiple principal components to achieve quantitative assessment of seedling quality, avoiding the one-sidedness of a single indicator; cluster analysis achieves automated grade classification, eliminating subjective human error and improving the consistency and reliability of grade classification; and by establishing differentiated control strategies, precise management of weak and vigorous seedlings is achieved. By standardizing agronomic knowledge and combining it with real-time data for parameter optimization, especially through response models of soil, weather, and measures, dynamic optimization of control measures is achieved, significantly improving the effects of promoting weak seedlings to become strong and controlling excessive growth. Therefore, the wheat overwintering seedling quality grade classification method and system proposed in this invention can solve the problem of low accuracy in wheat overwintering seedling quality grade classification. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating a method for classifying the quality grade of wheat seedlings during the overwintering period, as provided in an embodiment of the present invention. Figure 2 This is a functional module diagram of a wheat overwintering seedling quality grading system provided in an embodiment of the present invention.

[0009] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0010] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0011] This application provides a method for classifying the quality grade of wheat seedlings during the overwintering period. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for classifying the quality grade of wheat seedlings during the overwintering period can be executed by software or hardware installed on a terminal device or a server device. The software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0012] Reference Figure 1 The diagram shown is a flowchart illustrating a method for classifying the quality grade of wheat seedlings during the overwintering period according to an embodiment of the present invention. In this embodiment, the method for classifying the quality grade of wheat seedlings during the overwintering period includes: S1. Collect morphological and physiological-biochemical indicators of wheat seedlings during the wintering period, and use a preset multi-objective optimization model to analyze the dynamic thresholds corresponding to the morphological and physiological-biochemical indicators.

[0013] In this embodiment of the invention, the morphological and physiological-biochemical indicators of wheat seedlings during the overwintering period are collected. The morphological indicators include seedling height, leaf age, number of tillers per plant, number of secondary roots, and number of stem tillers per unit area. The physiological-biochemical indicators include chlorophyll content, soluble sugar content, antioxidant enzyme activity, dry weight of aboveground and underground parts, root activity, and root-to-shoot ratio of wheat leaves.

[0014] In detail, overwintering wheat seedlings refer to individual wheat seedlings that are in a low-temperature dormancy or slow growth stage from the time they cease growth in autumn until they resume growth in the following spring. Morphological indicators are parameters reflecting the external growth status of overwintering wheat seedlings, including seedling height, leaf age, number of tillers per plant, number of secondary roots, and number of stem tillers per unit area. Physiological and biochemical indicators are parameters reflecting the internal physiological metabolism and stress resistance of overwintering wheat seedlings, including chlorophyll content in functional leaves, soluble sugar content, antioxidant enzyme activity, aboveground and underground dry matter weight, root activity, and root-to-shoot ratio.

[0015] Specifically, a 1m×1m quadrat was delineated in the planting field using a five-point sampling method. Thirty seedlings were selected in each quadrat. Seedling height was measured with a ruler, leaf age was counted (one fully expanded leaf was counted, and 0.5 half-expanded leaves were counted), and the number of effective tillers per plant (≥2cm and with independent roots) was counted. The number of secondary rootings after digging up and rinsing the seedlings was counted, and the total number of stem tillers per unit area within the quadrat was counted. The newest fully expanded functional leaf of the main stem of the seedling was selected, and the SPAD value was measured using a chlorophyll meter. The soluble sugar content was measured using the anthrone colorimetric method, and the SOD activity was measured using the NBT photoreduction method. The aboveground and underground parts of the seedlings were separated, dried, weighed, and the dry matter weight was calculated. The root-to-shoot ratio was calculated, and the root activity was measured using the TTC reduction method.

[0016] Furthermore, morphological indicators and physiological and biochemical indicators have synergistic or restrictive relationships. If screening is based solely on a fixed threshold for a single indicator, this correlation will be ignored, resulting in screening results that cannot fully reflect the true growth status and stress resistance of seedlings. Therefore, it is necessary to analyze the dynamic thresholds for indicator screening, and dynamically adjust the screening criteria based on the actual importance of the indicators, model accuracy, and computational cost, to ensure that the selected indicators are both comprehensive and concise, meeting the evaluation needs of different scenarios.

[0017] In this embodiment of the invention, the dynamic threshold for index screening is obtained through multi-objective optimization model analysis and is adjusted according to the importance of the index, the accuracy of the model, and the cost. It is used as a critical value for screening key indicators.

[0018] In this embodiment of the invention, the step of using a preset multi-objective optimization model to analyze the dynamic threshold for index screening corresponding to the morphological and physiological / biochemical indicators includes: The importance score of each indicator among the morphological and physiological / biochemical indicators is calculated using the random forest algorithm. Each indicator is then sorted in descending order according to its importance score to generate a feature importance sequence. The model accuracy corresponding to each index in the feature importance sequence is analyzed using the model accuracy objective optimization function in the multi-objective optimization model. The target cost corresponding to each index in the feature importance sequence is analyzed using the cost objective optimization function in the multi-objective optimization model. A spatial decision boundary is generated based on the model accuracy and the target cost. The decision value corresponding to each indicator is analyzed based on the spatial decision boundary, and the dynamic threshold for indicator screening is determined based on the decision value.

[0019] In detail, the Random Forest algorithm is an ensemble learning algorithm that trains and predicts data by constructing multiple decision trees, and outputs the contribution of each feature to the model's prediction results. Collected morphological indicators (such as seedling height range of 8-15cm, number of tillers per plant of 1-3) and physiological and biochemical indicators (such as chlorophyll content of 35-50 SPAD value, soluble sugar content of 8-12mg / g) are used as input data and fed into the Random Forest algorithm. The algorithm generates multiple training sets through Bootstrap sampling, constructs a decision tree for each training set, and calculates the information gain or Gini coefficient reduction of each indicator during node splitting in each decision tree. This information gain is used as the importance score of that indicator. For example, if soluble sugar content has the highest information gain in the decision tree split (score 0.8), seedling height scores 0.6, and leaf age scores 0.4, then the features are sorted in descending order of score, generating a feature importance sequence of "soluble sugar content > seedling height > leaf age > number of tillers per plant > ...".

[0020] Specifically, the model accuracy objective optimization function is used in multi-objective optimization models to measure the model's accuracy in fitting or predicting indicator data. Each indicator in the feature importance sequence is sequentially incorporated into the model, and the accuracy of the model after incorporating that indicator is calculated using the model accuracy objective optimization function (e.g., mean squared error, MSE). For example, when soluble sugar content is included, the model MSE is 0.08; when soluble sugar content + seedling height is included, the MSE drops to 0.06; when all indicators are included, the MSE is 0.05, but the computational cost increases by 30%. The cost objective optimization function is used in multi-objective optimization models to measure the time, computing power, and other costs required during model computation. For each indicator in the feature importance sequence, the target cost after incorporating that indicator is analyzed using the cost objective optimization function (e.g., computation time, in seconds, ranging from 2-10 seconds). For example, when soluble sugar content is included, the computation time is 2 seconds; when root vitality (requiring laboratory testing data) is included, the computation time increases to 5 seconds because an additional data processing module needs to be called.

[0021] Furthermore, the spatial decision boundary is a critical curve in a two-dimensional space constructed based on model accuracy and target cost, used to delineate regions for inclusion or exclusion of indicators. With model accuracy (MSE) as the vertical axis and target cost (computation time) as the horizontal axis, the accuracy and cost data corresponding to each indicator are labeled in the two-dimensional space. The spatial decision boundary is generated through linear regression or support vector machine; for example, the region to the left of the boundary line is defined as "accuracy ≥ 0.95 and cost ≤ 6s," while the region to the right is defined as not meeting the criteria. Based on this boundary, the decision value corresponding to each indicator (such as the distance between the indicator and the boundary line) is analyzed. If the decision value is ≥ 0.7, the indicator is included in the screening range. Finally, a dynamic threshold for indicator screening is determined (such as importance score ≥ 0.5, accuracy improvement ≥ 0.03, and cost ≤ 5s), making indicator screening more aligned with practical application scenarios, balancing accuracy and practicality.

[0022] Furthermore, the dynamic threshold for index screening obtained through the multi-objective optimization model directly determines the screening criteria for key indicators.

[0023] S2. Dynamically filter the morphological indicators and the physiological and biochemical indicators according to the dynamic threshold of the indicators to obtain the key indicator dataset.

[0024] In this embodiment of the invention, the key indicator dataset is a set of indicators that are most representative for seedling quality assessment, which has been dynamically screened, subjected to collinearity analysis and standardized transformation.

[0025] In this embodiment of the invention, the step of dynamically filtering the morphological indicators and the physiological and biochemical indicators according to the indicator selection dynamic threshold to obtain a key indicator dataset includes: Based on the preset principal component loading analysis, the contribution of each of the morphological and physiological and biochemical indicators is calculated, and an importance ranking list of each indicator is generated according to the contribution. The critical point of contribution is determined by selecting dynamic thresholds based on the aforementioned indicators; The indicators are selected sequentially from the importance ranking list using a forward selection strategy until the critical point is reached; Multicollinearity analysis was performed on the selected indicators, and the indicator set after the multicollinearity analysis was standardized to generate a key indicator dataset.

[0026] In detail, the first step is to perform principal component loading analysis. Principal component loading analysis is a multivariate statistical analysis method that measures the contribution of an indicator to a principal component by calculating the correlation coefficient between the indicator and the principal component. Morphological indicators (such as seedling height 8-15cm, number of tillers per plant 1-3) and physiological and biochemical indicators (such as chlorophyll content 35-50SPAD value) are used as the analysis objects. The loading coefficient (i.e., correlation coefficient) of each indicator with each principal component is calculated through the covariance matrix. The larger the absolute value of the loading coefficient, the higher the contribution of the indicator to the principal component. For example, the loading coefficient of soluble sugar content with the first principal component is 0.9, seedling height is 0.7, and leaf age is 0.3. Then, an importance ranking list is generated in descending order of contribution: "soluble sugar content (0.9) > seedling height (0.7) > number of tillers per plant (0.6) > root activity (0.5) > ...". Convert the dynamic threshold (such as N) for index selection into a critical value corresponding to the contribution. For example, if N is 5, then the contribution of the 5th index is selected as the critical point. If the contribution of the 5th index is 0.5, then the critical point is the index with a contribution of 0.5.

[0027] Specifically, starting from the empty set, indicators are selected sequentially according to the importance ranking list: soluble sugar content (0.9) with the highest contribution is selected, and its contribution is judged to be ≥0.5 (critical point), and it is included in the set; then seedling height (0.7) is selected, which also meets the condition and is included in the set; then the number of tillers per plant (0.6) and root activity (0.5) are selected, both of which meet the condition; when the next indicator, leaf age (contribution 0.3) <0.5, the selection stops, and the selected indicator set at this time is "soluble sugar content, seedling height, number of tillers per plant, and root activity". Multicollinearity analysis was performed on the four selected indicators, using the variance inflation factor (VIF) as the criterion. A higher VIF value indicates stronger collinearity (typically, VIF > 10 indicates severe collinearity). For example, the calculated VIF for seedling height and number of tillers per plant was 8 (no severe collinearity). All indicators had VIF < 10 and did not need to be removed. If an indicator had a VIF = 12, it was removed, and the next indicator that met the contribution criteria was selected. The retained indicators were then standardized using the formula: ,in For standardized values, The original value, The mean, Using the standard deviation as a guideline, all indicators are now on the same order of magnitude after transformation.

[0028] Furthermore, the key indicator dataset is the direct input for principal component analysis. If the selected indicators have collinearity or dimensional differences, it will lead to deviations in the principal component calculation results. Therefore, it is necessary to avoid interference from invalid data.

[0029] S3. Perform principal component analysis on the key indicator dataset to obtain multiple principal components and their corresponding principal component scores.

[0030] In this embodiment of the invention, the principal component is a comprehensive index obtained after principal component analysis. Each principal component is a linear combination of the original indices. The first principal component contains the most variation information of the original data. The principal component score is the value of each sample on each principal component after substituting the original index data into the principal component linear expression.

[0031] In this embodiment of the invention, performing principal component analysis on the key indicator dataset to obtain multiple principal components and their corresponding principal component scores includes: Analyze the covariance matrix of the key indicator dataset, and perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; When the feature value is greater than a preset threshold, the index data corresponding to the feature value is used as the principal component; A linear function is constructed for each principal component based on the feature vector, and the principal component score value corresponding to each principal component is analyzed using the linear function.

[0032] In detail, principal component analysis is performed on agronomic traits such as relevant physiological and biochemical indicators, seedling height, leaf age, number of tillers per plant, aboveground dry matter weight, underground dry matter weight, root-to-shoot ratio, and number of secondary roots. This involves calculating the covariance matrix of the dataset, obtaining eigenvalues ​​and eigenvectors of the covariance matrix through eigenvalue decomposition, and selecting the comprehensive indicators corresponding to the first few eigenvalues ​​(the larger the eigenvalue, the more variation information of the original data contained in the corresponding principal component) based on the principle that the eigenvalue is greater than 1. Simultaneously, linear expressions for each principal component are constructed using the elements of the eigenvectors as coefficients (e.g., the first principal component = 0.4 × soluble sugar standardized value + 0.3 × seedling height standardized value + 0.2 × number of tillers per plant standardized value + 0.1 × root activity standardized value). When constructing the linear expressions for the principal components, each element in the eigenvector is used as a coefficient of the corresponding original key indicator. The coefficients are then summed sequentially ("coefficient × corresponding original key indicator standardized value") to obtain the linear expression for the principal component. For example, if the key indicator dataset contains four original key indicators: soluble sugar content (standardized as A), seedling height (standardized as B), number of tillers per plant (standardized as C), and root activity (standardized as D), after decomposing their covariance matrix eigenvalues, the eigenvector corresponding to the first principal component is [0.4, 0.3, 0.2, 0.1] (where 0.4 corresponds to soluble sugar content, 0.3 to seedling height, 0.2 to number of tillers per plant, and 0.1 to root activity), then the first principal component (denoted as D)... The linear expression of ) is If the eigenvector corresponding to the second principal component is [0.2, 0.1, 0.4, 0.3], then the second principal component (denoted as...) is... The linear expression of ) is .

[0033] Specifically, the standardized index value of each sample in the key indicator dataset is substituted into the linear expression of each principal component to calculate the value of each sample on the corresponding principal component. This value is the principal component score corresponding to each principal component, which realizes the dimensionality reduction of multi-dimensional key indicators. While retaining the core information of the original data, it simplifies subsequent analysis and provides concise and complete basic data for calculating the comprehensive value of wheat seedling quality during the wintering period.

[0034] S4. Standardize the principal component scores to obtain the comprehensive quality value of wheat seedlings during the wintering period. Perform cluster analysis on the wheat seedlings during the wintering period based on the comprehensive quality value to obtain the initial quality grade of the wheat seedlings during the wintering period.

[0035] In this embodiment of the invention, the comprehensive quality value refers to the quality value of wheat seedlings during the overwintering period.

[0036] In this embodiment of the invention, the standardization of the principal component scores to obtain the comprehensive quality value of wheat seedlings during the overwintering period includes: ; ; ; in, This represents the overall quality value of wheat seedlings during the wintering period. For the first Principal Components Corresponding standard data, The maximum value of the principal component score in each principal component. The minimum principal component score for each principal component. For the sample size, For the first The weight coefficients of each principal component For the first The eigenvalues ​​of each principal component.

[0037] In detail, using the formula Standardization processing is performed, among which For the first The scores of each principal component, The maximum value of the principal component score. The minimum value; for example, the score range of the first principal component is 0.2-1.0 ( , ), a certain sample ,but The score range for the second principal component is 0.1-0.8. , ),sample ,but Using formulas Calculate the weighting coefficients, for example, for the first principal component. Second principal component It can calculate the weight coefficients of different principal components and finally calculate the comprehensive quality value. By integrating information from multiple principal components, it solves the problem of incomplete evaluation by a single indicator in existing technologies. The technical effect is to obtain a quantitative value that reflects the overall quality of seedlings, providing an objective basis for grade classification.

[0038] In this embodiment of the invention, the initial quality level is obtained through cluster analysis, including vigorous seedlings, strong seedlings, and weak seedlings.

[0039] In this embodiment of the invention, the step of performing cluster analysis on wheat seedlings during the overwintering period based on the comprehensive quality value to obtain the initial quality grade of the wheat seedlings during the overwintering period includes: Pattern recognition was performed on the comprehensive quality values ​​to obtain the cluster centers and membership matrix of wheat seedlings during the wintering period; A cluster analysis diagram of wheat seedlings during the overwintering period is generated based on the cluster centers and the membership matrix, and the average membership value of the cluster analysis diagram is analyzed. The initial quality grade of wheat seedlings during the overwintering period is determined based on the comprehensive quality value and the average membership value, wherein the initial quality grade includes vigorous seedlings, strong seedlings and weak seedlings.

[0040] In detail, pattern recognition is performed on the overall quality value (e.g., sample D value range 0.3-0.9). Fuzzy C-means clustering algorithm is used, with a preset number of clusters of 3 (vigorous seedlings, strong seedlings, weak seedlings). The distance from each sample to each cluster is iteratively calculated to determine the cluster centers (e.g., vigorous seedling center D=0.85, cluster analysis result is level III; strong seedling center D=0.65, cluster analysis result is level II; weak seedling center D=0.45, cluster analysis result is level I) and the membership matrix (e.g., a sample has a membership degree of 0.8 for vigorous seedlings, 0.15 for strong seedlings, and 0.05 for weak seedlings). A cluster analysis graph is generated, with the overall quality value on the horizontal axis and the sample number on the vertical axis, plotted as a scatter plot. Different categories of samples are labeled with different colors to visually display the clustering results. Simultaneously, the average membership value of each category is calculated; for example, the average membership value of vigorous seedlings is 0.82; strong seedlings 0.75; and weak seedlings 0.78, all greater than 0.7, indicating good clustering results. Finally, the initial quality level is determined based on the comprehensive quality value and the average membership value: if D ≥ 0.65 and the cluster analysis result is Level III, it is classified as a vigorous seedling; if 0.40 ≤ D < 0.65 and the cluster analysis result is Level II, it is classified as a strong seedling; if D < 0.40 and the cluster analysis result is Level I, it is classified as a weak seedling.

[0041] Furthermore, the overall quality value is calculated based on the principal component score, and the initial quality level is a preliminary assessment result of the seedling quality; this initial level provides basic level data for spatiotemporal evolution analysis and provides a basis for subsequent optimization of the level.

[0042] S5. Based on the initial quality level and the preset climate change trend analysis, analyze the changing trend of wheat seedlings during the wintering period, optimize the initial quality level according to the changing trend to obtain the target quality level, and generate targeted control measures for wheat seedlings during the wintering period based on the target quality level.

[0043] In this embodiment of the invention, the trend of change refers to the direction and pattern of change in the quality grade of wheat seedlings during the wintering period within the target time period.

[0044] In this embodiment of the invention, the step of analyzing the changing trend of wheat seedlings during the overwintering period based on the initial quality level and the preset climate change trend includes: A spatiotemporal quality evolution dataset is constructed based on the initial quality level and the preset climate change trend; Seasonal decomposition is performed on the spatiotemporal quality evolution dataset to obtain the data evolution components; A spatiotemporal autoregressive analysis model for wheat seedlings during the overwintering period was constructed based on the trend and periodic components in the data evolution components. The spatiotemporal autoregressive analysis model was used to analyze the evolution trajectory of wheat seedling quality grades during the overwintering period, and the changing trend of wheat seedlings during the overwintering period was determined based on the evolution trajectory of quality grades.

[0045] In detail, the initial quality grades are aligned by timestamps and geographic coordinates to generate a standardized grade sequence. Temperature change rates and precipitation variability patterns from pre-defined climate change trends are extracted to form a climate impact factor set. The standardized grade sequence and the climate impact factor set are then fused to construct a multidimensional data cube. Missing value imputation and smoothing are performed on the multidimensional data cube to obtain the quality evolution dataset. The long-term trend in the quality evolution dataset is separated using the moving average method to obtain the trend component; Fourier transform is used to identify periodic fluctuations in the quality evolution dataset to extract the periodic component; the differences between the original data and the trend and periodic components are calculated to obtain the residual components. The significance of seasonal decomposition is verified through analysis of variance to ensure the effectiveness of component extraction.

[0046] Specifically, an autoregressive equation containing a spatial adjacency matrix and time lag terms is constructed, the model structure is defined, and the parameters of the autoregressive equation, including the spatial dependence coefficient and the time autoregressive coefficient, are solved using the maximum likelihood estimation method. The set of climate influencing factors is used as exogenous variables to enhance the explanatory power of the model. The model complexity is optimized through cross-validation to obtain a calibrated spatiotemporal autoregressive prediction model. The prediction time step and spatial range are set, the input data of the spatiotemporal autoregressive prediction model are initialized, the spatiotemporal autoregressive prediction model is iteratively run, a sequence of predicted values ​​for quality levels is generated, the confidence interval and uncertainty measure of the predicted value sequence are calculated, the prediction reliability is evaluated, the prediction results and uncertainty information are integrated, and a quantitative output of the changing trend is generated.

[0047] For example, by using a spatiotemporal autoregressive analysis model, the seedling quality level at each time point and spatial location is calculated and connected to form a quality level evolution trajectory. For instance, the seedlings in plot C were at level 2 in late November, dropping to level 1.5 in mid-December, then to level 1 in early January due to persistent low temperatures, and slowly recovering to level 1.2 in early February. Based on the common patterns of multiple trajectories, a final conclusion is reached. For example, under the pre-set trend of 2°C lower than normal temperature and 10% less than normal precipitation during the overwintering period, wheat seedlings initially at level 1-2 will experience a 20%-30% quality decrease in the mid-overwintering period (late December to early January), with the decrease being less pronounced in the southern regions than in the northern regions.

[0048] Furthermore, by observing changing trends, we can shift from passive response to proactive prediction, forecasting changes in crop conditions 15-30 days in advance, effectively identifying risk windows for meteorological disasters such as late spring frosts and droughts, and providing a scientific basis for taking preventive measures.

[0049] In this embodiment of the invention, the target quality grade is obtained by optimizing the initial quality grade, which is more in line with the quality grade of wheat seedlings during the wintering period.

[0050] In this embodiment of the invention, optimizing the initial quality level based on the changing trend to obtain the target quality level includes: Analyze the key driving factors in the changing trend, and generate the trend influence weight of the initial quality level based on the key driving factors; The classification boundary of the initial quality level is adjusted based on the trend influence weight, and an optimized classification rule for the initial quality level is generated based on the adjusted classification boundary. The optimized classification rules were used to analyze the preliminary target quality level of each sample in wheat seedlings during the overwintering period; The biological rationality and agronomic consistency of the preliminary target quality grade are verified to obtain the target quality grade.

[0051] In detail, gradient change information and inflection point features in the trend are extracted to identify key driving factors. If the trend shows that for every 5 days increase in the duration of low temperature during the overwintering period, the seedling quality grade decreases by 1 grade, and for every 10% decrease in soil moisture content, the grade decreases by 0.5 grades, then these two factors are the key driving factors. Fluctuations in sunshine duration have a negligible impact on the grade and are therefore excluded. The correlation coefficients between each driving factor and the change in quality grade are calculated to obtain the factor importance ranking. A judgment matrix is ​​constructed using the analytic hierarchy process (AHP) to quantify the relative importance among factors. The judgment matrix is ​​solved based on the eigenvector method to generate trend influence weights. The classification threshold intervals corresponding to the initial quality grades are obtained, including upper and lower thresholds. A mapping relationship is established between the trend influence weights and the adjustment amount of the classification thresholds. The correction amounts of the upper and lower thresholds are calculated based on the mapping relationship, and the classification threshold intervals are updated using the correction amounts to form optimized classification rules.

[0052] For example, based on the weights of key driving factors, the classification boundary is reset, and combined with the influence of the low temperature duration weight of 0.6, the classification boundary of leaf integrity rate is adjusted to a dynamic standard: Initial rule: Grade 1 leaf integrity rate ≥ 90%; Optimized rule: If the low temperature duration during the overwintering period is ≤ 7 days, the Grade 1 leaf integrity rate is ≥ 85% (the boundary is slightly stricter because the low temperature has a small impact); If the low temperature duration is > 7 days, the Grade 1 leaf integrity rate is ≥ 75% (the boundary is relaxed because the low temperature has a large impact, to avoid misjudgment).

[0053] Specifically, the characteristic index values ​​of each sample are input into the optimized classification rules to calculate the new classification score. Based on the new classification score and the updated classification threshold range, a provisional quality level for each sample is determined. Spatial consistency checks are performed on the provisional quality levels to eliminate outliers. The provisional quality levels of all samples are then aggregated to generate a preliminary target quality level. The preliminary target quality level is compared with historical agronomic data to check the rationality of the level distribution. Domain experts are invited to review the critical samples to correct significantly deviating levels. Statistical significance tests are conducted to ensure the stability and repeatability of the level division, and the validated target quality level is output.

[0054] Furthermore, by improving the misjudgment rate through trend correction, we can effectively identify seedling conditions at the critical stage of grade transition, providing targeted objectives for precise intervention.

[0055] In this embodiment of the invention, the control measures refer to a precision agricultural management plan formulated based on wheat seedling quality grade diagnosis and climate change prediction, which takes differentiated measures for different quality grades (weak seedlings / strong seedlings / vigorous seedlings).

[0056] In this embodiment of the invention, the targeted control measures for generating wheat overwintering seedlings based on the target quality grade include: Identify weak seedling groups and vigorous seedling groups in the target quality level, and generate a set of differentiated control objects; The first type of basic regulation scheme is generated based on the nutritional fortification and growth promotion measures for the weak seedling population in the set of differentiated regulation objects; A second type of basic regulation scheme is generated based on the growth regulation and nutrition control measures of the vigorous seedling population in the set of differentiated regulation objects; By pre-setting real-time soil monitoring data and meteorological forecast information, the parameters of the first type of basic regulation scheme and the second type of basic regulation scheme are optimized to obtain the regulation instructions for promoting weak growth and controlling excessive growth. Based on the aforementioned instructions to promote the transformation of weak seedlings into strong ones and the aforementioned instructions to control excessive growth and slow down seedling development, targeted control measures are to be taken for wheat seedlings during the wintering period.

[0057] In detail, the weak seedling grade identifiers are extracted from the target quality grade to determine the spatial distribution characteristics of weak seedling samples. The vigorous seedling grade identifiers are identified within the target quality grade to obtain the clustering area information of vigorous seedling samples. Field distribution maps of weak and vigorous seedlings are generated based on spatial interpolation methods, displaying the spatial patterns of different grades. Combined with the agronomic management unit division results, a differentiated set of control objects containing spatial locations is generated. The agronomic rule base storing knowledge on promoting weak seedlings to become strong and controlling excessive growth is accessed. Query conditions are defined, and for weak seedling groups, nutrient fortification and growth promotion measures are queried to obtain the first type of basic control scheme. For vigorous seedling groups, growth regulation and nutrient control measures are queried to obtain the second type of basic control scheme. The compatibility of the first and second types of basic control schemes is verified to ensure no conflicts exist between the measures.

[0058] For example, in the first type of basic regulation scheme, in view of the weak root absorption capacity of weak seedlings, quick-acting fertilizers (such as potassium dihydrogen phosphate solution) are applied, and key elements such as nitrogen, phosphorus and potassium are quickly supplemented by foliar spraying or small amount of trench application; the growth environment of weak seedlings is improved, such as shallow cultivation and loosening of soil in compacted areas to promote root respiration, and covering weak seedlings in low-temperature areas with straw to keep them warm and moist and reduce low-temperature stress. With promotion as the core, the measures are mild and targeted, and avoid seedling stress caused by excessive intervention. In the second category of basic regulation programs, plant growth regulators (such as paclobutrazol) are used to control excessive growth of seedlings through foliar spraying, shorten internode length, thicken stems, and enhance frost resistance. Strict dosage control is required to avoid excessive inhibition. Nitrogen fertilizer application should be reduced or suspended (nitrogen fertilizer will aggravate excessive growth). Phosphorus and potassium fertilizers should be supplemented appropriately to enhance the stress resistance of roots and stems. At the same time, water should be controlled appropriately to avoid excessive soil moisture that promotes growth and inhibits excessive growth of seedlings. The core of the program is control, and the measures focus on controlling excessive growth rather than damaging seedlings, balancing growth rate and stress resistance.

[0059] Specifically, the nitrogen, phosphorus, and potassium content and soil moisture index are obtained from real-time soil monitoring data to assess soil supply capacity. Temperature change trends and precipitation probabilities are extracted from meteorological forecasts to predict future growth environments. A soil-meteorology-measure response model is established to simulate the regulatory effects of different parameter combinations. The optimal parameter settings are solved using a multi-objective optimization algorithm to achieve refined adjustments to the basic regulation scheme. Based on the optimized parameters, topdressing formulas and foliar fertilizer application sequences for weak seedlings are formulated, generating instructions to promote the growth of weak seedlings. Chemical regulation measures for vigorous seedlings are determined based on the type and concentration of growth regulators, generating instructions to control excessive growth and slow down excessive growth. The spatiotemporal execution order of the two types of instructions is coordinated to avoid mutual interference between measures. All regulation instructions are integrated into a unified field management scheme, with additional explanations of execution conditions. For example, in plot 1 (30% of seedlings are weak): Before December 15th (before rainfall), potassium dihydrogen phosphate is sprayed at a concentration of 0.2%, once every 3 days, for a total of 2 times; at the same time, it is covered with 5cm thick straw for key insulation.

[0060] Furthermore, by establishing differentiated control strategies, precise management of weak and vigorous seedlings was achieved. By standardizing agronomic knowledge and optimizing parameters using real-time data, and through a soil-meteorological-measure response model, dynamic optimization of control measures was realized, significantly improving the effects of promoting the transformation of weak seedlings into strong ones and controlling excessive growth.

[0061] like Figure 2 The diagram shown is a functional block diagram of a wheat overwintering seedling quality grading system provided in an embodiment of the present invention.

[0062] The wheat overwintering seedling quality grading system 100 of this invention can be installed in an electronic device. Depending on the functions implemented, the wheat overwintering seedling quality grading system 100 may include an indicator screening dynamic threshold analysis module 101, a key indicator dataset generation module 102, a principal component analysis module 103, an initial quality grade analysis module 104, and a control measure generation module 105. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0063] In this embodiment, the functions of each module / unit are as follows: The indicator screening dynamic threshold analysis module 101 is used to collect morphological and physiological and biochemical indicators of wheat seedlings during the wintering period, and to analyze the indicator screening dynamic thresholds corresponding to the morphological and physiological and biochemical indicators using a preset multi-objective optimization model. The key indicator dataset generation module 102 is used to dynamically filter the morphological indicators and the physiological and biochemical indicators according to the indicator screening dynamic threshold to obtain the key indicator dataset. The principal component analysis module 103 is used to perform principal component analysis on the key indicator dataset to obtain multiple principal components and the principal component score values ​​corresponding to the multiple principal components. The initial quality grade analysis module 104 is used to standardize the principal component scores to obtain the comprehensive quality value of wheat overwintering seedlings, and to perform cluster analysis on the wheat overwintering seedlings based on the comprehensive quality value to obtain the initial quality grade of the wheat overwintering seedlings. The regulation measure generation module 105 is used to analyze the changing trend of wheat seedlings during the wintering period based on the initial quality level and the preset climate change trend, optimize the initial quality level according to the changing trend to obtain the target quality level, and generate targeted regulation measures for wheat seedlings during the wintering period based on the target quality level.

[0064] In detail, the modules in the wheat overwintering seedling quality grading system 100 described in this embodiment of the invention adopt the same approach as described above when in use. Figure 1 The method used is the same as the method for classifying the quality grade of wheat seedlings during the wintering period described in the article, and can produce the same technical effect, so it will not be repeated here.

[0065] In the several embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0066] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0067] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0068] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0069] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects. The scope of the invention is not limited to the foregoing description, and all variations within the meaning and scope of equivalents falling within the protection scope are intended to be included in the invention.

[0070] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0071] Furthermore, it is clear that the word "including" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems described in a system item may also be implemented by a single unit or system through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for classifying the quality of wintering seedlings of wheat, characterized by, The method comprises: Collecting morphological indexes and physiological and biochemical indexes of wintering wheat seedlings, and analyzing index screening dynamic thresholds corresponding to the morphological indexes and the physiological and biochemical indexes by using a preset multi-objective optimization model; Performing dynamic screening on the morphological indexes and the physiological and biochemical indexes according to the index screening dynamic thresholds to obtain a key index data set; Performing principal component analysis on the key index data set to obtain a plurality of principal components and principal component score values corresponding to the plurality of principal components; Performing standardization processing on the principal component score values to obtain a quality comprehensive value of the wintering wheat seedlings, performing cluster analysis on the wintering wheat seedlings based on the quality comprehensive value, and obtaining an initial quality grade of the wintering wheat seedlings; Analyzing a change trend of the wintering wheat seedlings based on the initial quality grade and a preset climate change trend, optimizing the initial quality grade according to the change trend, obtaining a target quality grade, and generating targeted regulation and control measures for the wintering wheat seedlings based on the target quality grade.

2. The method for classifying the quality of wintering seedlings of wheat according to claim 1, characterized by, The collecting of the morphological indexes and the physiological and biochemical indexes of the wintering wheat seedlings, wherein the morphological indexes include seedling height, leaf age, number of tillers per plant, number of secondary roots, and number of stem tillers per unit area; and the physiological and biochemical indexes include chlorophyll content, soluble sugar content, antioxidant enzyme activity, aboveground and underground dry matter weight, root activity, and root-shoot ratio.

3. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The analyzing of the index screening dynamic thresholds corresponding to the morphological indexes and the physiological and biochemical indexes by using the preset multi-objective optimization model comprises: Calculating importance scores of each index in the morphological indexes and the physiological and biochemical indexes by using a random forest algorithm, arranging each index in descending order according to the importance scores, and generating a feature importance sequence; Analyzing model precision corresponding to each index in the feature importance sequence by using a model precision objective optimization function in the multi-objective optimization model; Analyzing a target cost corresponding to each index in the feature importance sequence by using a cost objective optimization function in the multi-objective optimization model; Generating a spatial decision boundary based on the model precision and the target cost, analyzing a decision value corresponding to each index based on the spatial decision boundary, and determining an index screening dynamic threshold according to the decision value.

4. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The dynamic screening of the morphological indexes and the physiological and biochemical indexes according to the index screening dynamic threshold to obtain a key index data set comprises: Calculating contribution degrees of each index in the morphological indexes and the physiological and biochemical indexes based on preset principal component loading analysis, and generating an importance sorting list of each index according to the contribution degrees; Determining a critical point of the contribution degrees according to the index screening dynamic threshold; Selecting indexes from the importance sorting list in turn by using a forward selection strategy until the critical point is reached; Performing multicollinearity analysis on the selected indexes, and generating a key index data set by standardizing and converting the index set that has undergone the multicollinearity analysis.

5. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The standardization processing of the principal component score values to obtain a quality comprehensive value of the wintering wheat seedlings comprises: ； ； ； in, This represents the overall quality value of wheat seedlings during the wintering period. For the first Principal Components Corresponding standard data, The maximum value of the principal component score in each principal component. The minimum principal component score for each principal component. For the sample size, For the first The weight coefficients of each principal component For the first The eigenvalues ​​of each principal component.

6. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The clustering analysis of the wheat overwintering seedlings is performed based on the quality comprehensive value, and an initial quality grade of the wheat overwintering seedlings is obtained, including: The quality comprehensive value is subjected to pattern recognition, and a clustering center and a membership matrix of the wheat overwintering seedlings are obtained; A clustering analysis graph of the wheat overwintering seedlings is generated according to the clustering center and the membership matrix, and an average membership value result of the clustering analysis graph is analyzed; The initial quality grade of the wheat overwintering seedlings is determined according to the quality comprehensive value and the average membership value result, wherein the initial quality grade includes vigorous seedlings, strong seedlings and weak seedlings.

7. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The change trend of the wheat overwintering seedlings is analyzed based on the initial quality grade and a preset climate change trend, including: A spatiotemporal quality evolution dataset is constructed according to the initial quality grade and the preset climate change trend; The spatiotemporal quality evolution dataset is subjected to seasonal decomposition, and data evolution components are obtained; A spatiotemporal autoregressive analysis model of the wheat overwintering seedlings is constructed based on trend components and periodic components in the data evolution components; The quality grade evolution trajectory of the wheat overwintering seedlings is analyzed through the spatiotemporal autoregressive analysis model, and the change trend of the wheat overwintering seedlings is determined based on the quality grade evolution trajectory.

8. The method for classifying the quality of wintering seedlings of wheat according to claim 1, wherein The initial quality grade is optimized according to the change trend, and a target quality grade is obtained, including: Key driving factors in the change trend are analyzed, and a trend influence weight of the initial quality grade is generated according to the key driving factors; The classification boundary of the initial quality grade is adjusted based on the trend influence weight, and an optimized classification rule of the initial quality grade is generated based on the adjusted classification boundary; The preliminary target quality grade of each sample in the wheat overwintering seedlings is analyzed using the optimized classification rule; The target quality grade is obtained by verifying the preliminary target quality grade for biological rationality and agronomic consistency.

9. The wheat overwintering seedling quality grade classification method of claim 1, wherein: Targeted regulation measures for the wheat overwintering seedlings are generated based on the target quality grade, including: Weak seedling groups and vigorous seedling groups in the target quality grade are identified, and a differentiated regulation object set is generated; A first type of basic regulation scheme is generated according to nutrient enrichment and growth promotion measures for the weak seedling groups in the differentiated regulation object set; A second type of basic regulation scheme is generated according to growth regulation and nutrient control measures for the vigorous seedling groups in the differentiated regulation object set; Parameter optimization is performed on the first type of basic regulation scheme and the second type of basic regulation scheme through preset real-time soil monitoring data and weather forecast information, and weak-to-strong conversion regulation instructions and vigorous growth control regulation instructions are obtained; Targeted regulation measures for the wheat overwintering seedlings are determined according to the weak-to-strong conversion regulation instructions and the vigorous growth control regulation instructions.

10. A system for classifying the quality of wintering seedlings of wheat, characterized by A system for performing the wheat overwintering seedling quality grade classification method of any one of claims 1-9, the system comprising: The index screening dynamic threshold analysis module is configured to collect morphological indexes and physiological and biochemical indexes of the wheat seedlings in the overwintering period, and analyze index screening dynamic thresholds corresponding to the morphological indexes and the physiological and biochemical indexes by using a preset multi-objective optimization model. The key index data set generation module is configured to dynamically screen the morphological indexes and the physiological and biochemical indexes according to the index screening dynamic thresholds, and obtain a key index data set. The principal component analysis module is configured to perform principal component analysis on the key index data set, and obtain a plurality of principal components and principal component score values corresponding to the plurality of principal components. The initial quality grade analysis module is configured to perform standardization processing on the principal component score values, obtain a quality comprehensive value of the wheat seedlings in the overwintering period, perform cluster analysis on the wheat seedlings in the overwintering period based on the quality comprehensive value, and obtain an initial quality grade of the wheat seedlings in the overwintering period. The regulation measure generation module is configured to analyze a change trend of the wheat seedlings in the overwintering period based on the initial quality grade and a preset climate change trend, optimize the initial quality grade according to the change trend, obtain a target quality grade, and generate targeted regulation measures for the wheat seedlings in the overwintering period based on the target quality grade.

Images