Photosynthetic dynamic regulation and control prediction method based on bleeding sap concentration time sequence change

Abstract

The invention relates to the field of agricultural data analysis, in particular to a bleeding sap concentration time sequence change-based photosynthetic dynamic regulation and control prediction method, which comprises the following steps of: preprocessing sucrose concentration and abscisic acid concentration data of corn bleeding sap to obtain smooth time sequence physiological status data; performing dynamic minimum value analysis on the preprocessed physiological state data to obtain a membership function of the compensatory response region; performing joint evaluation on the change rate and the risk level of the physiological state point to obtain an adaptive distance regulation factor; an adaptive distance adjustment factor is introduced into Euclidean distance calculation, and optimized distance measurement is obtained and clustering diagnosis is carried out; and performing state mapping on a clustering diagnosis result to obtain a photosynthetic dynamic prediction conclusion and a regulation and control strategy so as to solve the problem of misjudgment of photosynthetic dynamic prediction caused by incapability of correctly identifying a corn compensatory physiological regulation state in the prior art.

Description

Technical Field

[0001] This invention relates to the field of agricultural data analysis technology, and in particular to a method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration. Background Technology

[0002] In the development of modern precision agriculture and smart agriculture, how to achieve real-time perception and dynamic regulation of crop physiological state has become a key issue in improving crop yield and quality. Photosynthesis, as the core link in crop growth and nutrient accumulation, directly determines the crop's growth rate and its responsiveness to environmental conditions through its dynamic changes. Therefore, accurate prediction of photosynthetic dynamics and the formulation of reasonable regulation strategies based on this are important directions for realizing intelligent agricultural production. Plant sap, as an important medium for long-distance transport of water and nutrients within plants, can sensitively reflect the crop's photosynthetic level and stress response state through temporal changes in its chemical composition. Sucrose concentration characterizes the accumulation level of photosynthetic products, while abscisic acid concentration is closely related to stomatal regulation and stress resistance mechanisms. With the advancement of sensing technology, by deploying sap sensors and environmental sensor networks in the field, it is possible to continuously collect multi-dimensional data such as sucrose concentration, abscisic acid concentration, light intensity, and air temperature and humidity, providing an important data source for photosynthetic dynamic analysis based on physiological characteristics.

[0003] In existing technologies, unsupervised clustering methods based on distance metrics, such as the DBSCAN clustering algorithm, are commonly used to diagnose crop status from complex multidimensional physiological time-series data. These methods classify physiological state points into normal and abnormal states by calculating the distance relationships between them in the feature space, thus providing initial conditions for photosynthetic dynamics prediction models. However, traditional methods mainly rely on standard Euclidean distance to directly accumulate numerical differences, which can easily lead to misjudgments when crops are in certain special physiological regulatory states. For example, when corn experiences a sudden shift from sunny to cloudy weather, it actively reduces its abscisic acid concentration to maintain high stomatal conductance and improve photosynthetic efficiency under low light. This compensatory regulatory behavior is numerically manifested as a decrease in sucrose concentration and a significant drop in abscisic acid concentration, which differs considerably from the normal sunny state. Because Euclidean distance cannot recognize this physiologically reasonable regulation, this state point is often incorrectly judged as abnormal, even considered more abnormal than states caused by disease or nutrient stress, leading to deviations in subsequent prediction and control strategies. Therefore, existing technologies have the problem of failing to correctly identify compensatory regulatory states when predicting photosynthetic dynamics in crops with high light efficiency and high transpiration, such as maize. There is an urgent need to propose an improved method that can combine the essential physiological characteristics with the background of environmental changes in order to improve the accuracy and reliability of photosynthetic dynamics prediction and regulation. Summary of the Invention

[0004] In view of this, the present invention aims to propose a method for predicting photosynthetic dynamic regulation based on the temporal changes in sap concentration, so as to solve the problem of misjudgment of photosynthetic dynamic prediction caused by the inability to correctly identify the compensatory physiological regulatory state of maize in the prior art.

[0005] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0006] A method for predicting photosynthetic dynamic regulation based on temporal changes in sap fluid concentration, comprising:

[0007] Step S1: Obtain smooth time-series physiological state data by preprocessing the sucrose and abscisic acid concentration data of corn sap.

[0008] Step S2: Obtain the membership function of the compensatory response region by performing dynamic minimum analysis on the preprocessed physiological state data;

[0009] Step S3: Obtain the adaptive distance adjustment factor by jointly assessing the rate of change of physiological state points and risk level;

[0010] Step S4: By introducing an adaptive distance adjustment factor into the Euclidean distance calculation, the optimized distance metric is obtained and cluster diagnosis is performed;

[0011] Step S5: By performing state mapping on the cluster diagnosis results, obtain the dynamic prediction conclusions and regulation strategies for photosynthesis.

[0012] Furthermore, the process of preprocessing the sucrose and abscisic acid concentration data in corn sap to obtain smooth time-series physiological state data includes:

[0013] Set the sensor sampling frequency; deploy sap flow sensors in the corn growing environment and continuously collect time-series data on sucrose and abscisic acid concentrations in corn stalk sap according to the sensor sampling frequency; deploy an environmental sensor network in the corn growing environment to synchronously collect environmental parameter data, including light intensity data, air temperature data, and air humidity data; input the collected raw time-series data into a preprocessing process, which includes: using... The criteria involve scanning the time-series data of sucrose concentration and abscisic acid concentration, identifying and removing outliers; filling in missing data points after outlier removal using linear interpolation; and smoothing the corrected time-series data of sucrose concentration and abscisic acid concentration using moving average filtering to obtain preprocessed time-series data of sucrose concentration and abscisic acid concentration.

[0014] Furthermore, the step of obtaining the membership function of the compensatory response region by performing dynamic minimum value analysis on the preprocessed physiological state data includes:

[0015] By setting a retrospective time window length and performing retrospective time window analysis on sucrose concentration data and abscisic acid concentration data, the characteristic center value of the compensatory response region is obtained.

[0016] For any target time point in the sucrose concentration time series data and abscisic acid concentration time series data, the membership assessment of the physiological state data at the target time point to the compensatory response region is obtained by performing variance evaluation and normalization on the sucrose concentration data and abscisic acid concentration data at the target time point based on the characteristic central value of the compensatory response region.

[0017] Furthermore, the step of obtaining the characteristic center value of the compensatory response region by performing a retrospective time window analysis on sucrose concentration data and abscisic acid concentration data includes:

[0018] For any target time in the time-series physiological state data, preprocessed sucrose concentration data from consecutive sampling times are selected within the retrospective time window corresponding to the target time. The minimum value of the sucrose concentration data within this time window is obtained, and this minimum value is used as the feature center value of the compensatory response region in the sucrose concentration dimension. Preprocessed abscisic acid concentration data from consecutive sampling times are selected within the target time window. The minimum value of the abscisic acid concentration data within this time window is obtained, and this minimum value is used as the feature center value of the compensatory response region in the abscisic acid concentration dimension.

[0019] Furthermore, the step of obtaining the membership assessment of the physiological state data at the target time to the compensatory response region by performing variance evaluation and normalization on the sucrose concentration data and abscisic acid concentration data at the target time based on the characteristic center value of the compensatory response region includes:

[0020] The variance of sucrose concentration data in the retrospective time window corresponding to the target time is used as the range parameter for the sucrose concentration dimension; the variance of abscisic acid concentration data in the retrospective time window corresponding to the target time is used as the range parameter for the abscisic acid concentration dimension.

[0021] The numerator is the result of subtracting the feature center value of the compensatory response region at the target time from the sucrose concentration data at the target time in the sucrose concentration dimension. The denominator is the square of the range parameter of the sucrose concentration dimension at the target time. The resulting fraction is used as the first evaluation factor of the sucrose concentration at the target time.

[0022] The numerator is the result of subtracting the feature center value of the compensatory response region in the abscisic acid concentration dimension from the abscisic acid concentration data at the target time, and the denominator is the square of the range parameter of the abscisic acid concentration dimension at the target time. The resulting fraction is used as the first evaluation factor of the abscisic acid concentration at the target time.

[0023] The negative of the result of subtracting the first evaluation factor of the sucrose concentration at the target time from the first evaluation factor of the abscisic acid concentration at the target time is used for exponential mapping with the natural constant as the base. The corresponding mapping result is used as the membership assessment of the physiological state data at the target time to the compensatory response region.

[0024] Furthermore, the method of obtaining an adaptive distance adjustment factor by jointly assessing the rate of change of physiological state data and risk level includes:

[0025] By conducting risk level analysis on sucrose concentration data and abscisic acid concentration data, penalty factors were obtained.

[0026] Reward factors were obtained by performing data homogeneity analysis on sucrose concentration data and abscisic acid concentration data;

[0027] By jointly normalizing the penalty factor and the reward factor, an adaptive distance adjustment factor for differentiated distance adjustment is obtained.

[0028] Furthermore, the step of obtaining a penalty factor through risk level analysis of sucrose concentration data and abscisic acid concentration data includes:

[0029] At any target time, the rate of change of sucrose concentration at that time is obtained by subtracting the sucrose concentration at the adjacent sampling point at the next time from the sucrose concentration at the previous time from the sucrose concentration at the next time, and dividing the result by the sampling time interval. The rate of change of abscisic acid concentration at that time is also obtained by subtracting the abscisic acid concentration at the adjacent sampling point at the next time from the abscisic acid concentration at the previous time from the abscisic acid concentration at the next time. The rate of change of sucrose concentration and the rate of change of abscisic acid concentration are then vector-synthesized to obtain the rate of change vector of physiological state data at the target time.

[0030] For any target time, the Mahalanobis distance between the physiological state data at the target time and the average physiological state data in the corresponding retrospective time window is used as the first risk assessment for the target time; for any two target times, the first target time and the second target time, the mean of the first risk assessment at the first target time and the first risk assessment at the second target time is used as the second risk assessment between the first target time and the second target time; the second risk assessment is mapped using the hyperbolic tangent function, and the corresponding mapping result is used as the third risk assessment between the first target time and the second target time;

[0031] The average vector of the rate of change vectors of physiological state data at the first target time and the rate of change vectors of physiological state data at the second target time is used as the average rate of change vector at the first target time and the second target time. The square of the magnitude of the vector obtained by subtracting the rate of change vector of physiological state data at the first target time from the corresponding average rate of change vector is used as the co-evaluation at the first target time. The square of the magnitude of the vector obtained by subtracting the rate of change vector of physiological state data at the second target time from the corresponding average rate of change vector is used as the co-evaluation at the second target time.

[0032] The numerator is the result of adding the collaborative evaluation at the first target time and the collaborative evaluation at the second target time. The denominator is the result of adding the square of the magnitude of the rate of change vector of the physiological state data at the first target time and the square of the magnitude of the rate of change vector of the physiological state data at the second target time. The corresponding fraction is used as the collaborative evaluation between the first target time and the second target time.

[0033] The result of multiplying the third risk assessment and synergy assessment at the first and second target times is used as the penalty factor at the first and second target times.

[0034] Furthermore, the step of obtaining the reward factor by performing data homogeneity analysis on sucrose concentration data and abscisic acid concentration data includes:

[0035] For any two target times, the first target time and the second target time, the membership evaluation of the physiological state data of the first target time to the compensatory response region and the membership evaluation of the physiological state data of the second target time to the compensatory response region are multiplied and the square root is calculated. The corresponding calculation result is used as the joint evaluation of the membership of the first target time and the second target time.

[0036] The constant 1 minus the calculation result of the joint evaluation of the membership degree of the first target time and the second target time is used as the gating factor of the first target time and the second target time.

[0037] The target time with the later timestamp between the first target time and the second target time is taken as the baseline target time. The average distance between all physiological state data in the retrospective time window corresponding to the baseline target time and the mean point of all physiological state data in the retrospective time window is taken as the basic distance constant between the first target time and the second target time.

[0038] The Euclidean distance between the physiological state data corresponding to the first target time and the second target time is used as the basic Euclidean distance between the first target time and the second target time.

[0039] The square of the basic distance constant between the first target time and the second target time is used as the numerator, and the sum of the square of the basic distance constant between the first target time and the second target time and the square of the basic Euclidean distance is used as the denominator. The corresponding fraction is used as the first reward evaluation for the first target time and the second target time.

[0040] The result of multiplying the first risk assessment at the first target time and the first risk assessment at the second target time and taking the square root is used as the comprehensive risk assessment at the first target time and the second target time; the result of adding the comprehensive risk assessment at the first target time and the second target time to the constant 1 as the numerator and the constant 1 as the denominator is used as the corresponding fraction as the second reward assessment at the first target time and the second target time.

[0041] The calculation result of multiplying the gating factors of the first target time and the second target time, and the first reward evaluation and the second reward evaluation, is used as the reward factors of the first target time and the second target time.

[0042] Furthermore, the step of obtaining an adaptive distance adjustment factor for differentiated distance adjustment by jointly normalizing the rate of change vector and the risk level index includes:

[0043] For any two target times, the first target time and the second target time, the result of adding the penalty factor of the first target time and the constant 1 to the second target time is used as the numerator, and the result of adding the reward factor of the first target time and the constant 1 to the second target time is used as the denominator. The corresponding fraction is used as the adaptive distance adjustment factor of the first target time and the second target time.

[0044] Furthermore, the step of introducing an adaptive distance adjustment factor into the Euclidean distance calculation to obtain an optimized distance metric and perform cluster diagnosis includes:

[0045] For any two target times, the first target time and the second target time, the square root of the product of the square of the basic Euclidean distance between the first target time and the second target time and the adaptive distance adjustment factor is used as the optimized distance metric between the first target time and the second target time. Cluster analysis is performed using the optimized distance metric between the first target time and the second target time to obtain the cluster diagnosis results.

[0046] Compared with the prior art, the present invention has the following advantages:

[0047] The photosynthetic dynamic regulation prediction method based on the temporal variation of sap concentration described in this invention, by introducing a membership function and an adaptive distance adjustment factor, differentiates and optimizes the physiological state of maize under different light conditions, effectively overcoming the misjudgment problem caused by traditional Euclidean distance relying solely on numerical differences. Under continuous cloudy or low-light conditions, maize often actively reduces abscisic acid concentration to improve stomatal conductance, a healthy compensatory regulatory process, but which is often incorrectly identified as an abnormal state in existing methods. This invention accurately captures and quantifies this special regulatory behavior by constructing a membership function dynamically correlated with the physiological lower limit, thereby avoiding the misreporting of healthy regulatory states as abnormal points and ensuring accurate diagnosis of the true physiological state of maize. In practical applications, this invention incorporates the rate of change of physiological state points and risk levels into a joint assessment, and achieves differentiated processing of different state points through the action of the adaptive distance adjustment factor. For true stress conditions, this invention significantly amplifies the distance from normal conditions, thereby improving the sensitivity of anomaly identification. For normal diurnal variation conditions, a reward mechanism enhances cluster cohesion, making normal data clusters clearer. For compensatory regulation conditions, neutrality is maintained to avoid false alarms and missed alarms. This design not only improves the accuracy of photosynthetic dynamics prediction but also provides reliable decision-making basis for automated control systems, reducing unnecessary interventions and misleading operations, thereby achieving precise control of maize growth and improving yield and quality. Attached Figure Description

[0048] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0049] Figure 1 This is a flowchart of the photosynthetic dynamic regulation prediction method based on the temporal change of sap concentration, as described in an embodiment of the present invention. Detailed Implementation

[0050] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0051] See Figure 1 This is a flowchart of the photosynthetic dynamic regulation prediction method based on the temporal change of sap concentration provided in Embodiment 1 of the present invention, as follows: Figure 1 As shown, the photosynthetic dynamic regulation prediction method based on the temporal variation of sap concentration can include:

[0052] Step S1: By preprocessing the data on sucrose concentration and abscisic acid concentration in corn sap, smooth time-series physiological state data are obtained.

[0053] First, the sensor sampling frequency is set; in this embodiment, the sampling frequency is set to once every 5 minutes. A sap flow sensor is deployed in the corn growing environment, and time-series data of sucrose concentration and abscisic acid concentration in the corn stalk sap are continuously collected according to the sensor sampling frequency. An environmental sensor network is deployed in the corn growing environment to synchronously collect environmental parameter data, including light intensity data, air temperature data, and air humidity data. The collected raw time-series data is input into a preprocessing flow, which includes: using... The criteria involve scanning the time-series data of sucrose concentration and abscisic acid concentration, identifying and removing outliers; filling in missing data points after outlier removal using linear interpolation; and smoothing the corrected time-series data of sucrose concentration and abscisic acid concentration using moving average filtering to obtain preprocessed time-series data of sucrose concentration and abscisic acid concentration.

[0054] Thus, the process of preprocessing data on sucrose and abscisic acid concentrations in corn sap to obtain smooth time-series physiological state data has been completed.

[0055] Step S2: By performing dynamic minimum analysis on the preprocessed physiological state data, the membership function of the compensatory response region is obtained.

[0056] The core problem this invention addresses is that standard Euclidean distance cannot identify the compensatory physiological regulation exhibited by maize when transitioning from prolonged sunny to persistently cloudy weather. This state essentially involves maize actively reducing its abscisic acid (ABA) concentration to maximize stomatal conductance in response to low light conditions—an adaptive strategy. However, the data shows a significant difference between this state point (low sucrose concentration, extremely low ABA concentration) and the normal sunny state point (high sucrose concentration, moderate ABA concentration), leading Euclidean distance to incorrectly classify it as abnormal. Therefore, the first optimization step of this invention aims to preliminarily determine whether a state point falls within the compensatory response region defined by (low sucrose concentration, extremely low ABA concentration). This step achieves this by constructing a membership function, which calculates the degree to which any state point belongs to a dynamically determined compensatory response region. The center and extent of this region are adaptively determined through statistical analysis of recently collected (past 24 hours) data. The output value of this membership function serves as a foundational signal, providing a basis for subsequent distance optimization adjustments.

[0057] In summary, the length of the retrospective time window is first set. In this embodiment of the invention, the Silicon Valley trip time window is set to 24 hours. By performing retrospective time window analysis on sucrose concentration data and abscisic acid concentration data, the feature center value of the compensatory response region is obtained. Specifically, for any target time in the time-series physiological state data, preprocessed sucrose concentration data from consecutive sampling times are selected within the retrospective time window corresponding to the target time. The minimum value of the sucrose concentration data within this time window is obtained and used as the feature center value of the compensatory response region in the sucrose concentration dimension. Similarly, preprocessed abscisic acid concentration data from consecutive sampling times are selected within the target time window. The minimum value of the abscisic acid concentration data within this time window is obtained and used as the feature center value of the compensatory response region in the abscisic acid concentration dimension.

[0058] After obtaining the feature center values ​​of the compensatory response region, for any target time point in the sucrose concentration time series data and abscisic acid concentration time series data, variance evaluation and normalization are performed on the sucrose concentration data and abscisic acid concentration data at the target time point based on the feature center values ​​of the compensatory response region. This yields an evaluation of the membership of the physiological state data at the target time point to the compensatory response region. Specifically, the variance of the sucrose concentration data within the retrospective time window corresponding to the target time point is used as the range parameter for the sucrose concentration dimension; the variance of the abscisic acid concentration data within the retrospective time window corresponding to the target time point is used as the range parameter for the abscisic acid concentration dimension; and the result of subtracting the feature center value of the compensatory response region at the target time point from the sucrose concentration data at the target time point is calculated. As the numerator, the square of the range parameter of the sucrose concentration dimension at the target time is used as the denominator, and the corresponding fraction is used as the first evaluation factor of the sucrose concentration at the target time. The result of subtracting the abscisic acid concentration data at the target time from the feature center value of the compensatory response region at the target time in the abscisic acid concentration dimension is used as the numerator, and the square of the range parameter of the abscisic acid concentration dimension at the target time is used as the denominator, and the corresponding fraction is used as the first evaluation factor of the abscisic acid concentration at the target time. The negative of the result of subtracting the first evaluation factor of the sucrose concentration at the target time from the first evaluation factor of the abscisic acid concentration at the target time is used for exponential mapping with the natural constant as the base, and the corresponding mapping result is used as the membership evaluation of the physiological state data at the target time to the compensatory response region.

[0059] In one implementation, assume the first The sucrose concentration at that time was ;No. The abscisic acid concentration at that time was ;No. The feature center value of the sucrose concentration dimension at time t is ;No. The eigencenter value of the abscisic acid concentration dimension at time t is The range parameter for the sucrose concentration dimension is: The range parameter for abscisic acid concentration is: Then the first The formula for calculating the membership evaluation of the compensatory response region based on the physiological state data at each time point is as follows:

[0060]

[0061] in, Indicates the first Evaluation of the membership of physiological state data at each time point to the compensatory response region; Indicates the first The sucrose concentration at a given moment; Indicates the first The feature center value of the sucrose concentration dimension at each time point; A range parameter representing the dimension of sucrose concentration; Indicates the first Abscisic acid concentration at each moment; Indicates the first The characteristic center value of the abscisic acid concentration dimension at each time point; A range parameter indicating the concentration dimension of abscisic acid; Let e ​​represent the natural constant.

[0062] It should be noted that this membership function transforms the concentration value of a state point into its proximity to a dynamically determined compensatory response feature center, outputting a value within the range of... The membership value of the interval. When the state point is exactly located at the center, the membership evaluation value is 1; as the state data point moves away from the center, the membership evaluation value smoothly decays to 0. In the context of continuous cloudy weather, the compensatory regulation of maize health is physiologically characterized by abscisic acid and sucrose concentrations approaching or falling below their physiological lower limits in the normal diurnal variation cycle. The reason why existing Euclidean distance methods produce misjudgments is precisely because they cannot identify this specific health region. This invention defines the characteristic center of the membership evaluation directly as the minimum value in recent (past 24 hours) observation data. The closer the concentration value of a test physiological state data point is to this physiological lower limit center defined by real data, the closer its calculated membership evaluation is to 1. In this way, the membership evaluation function no longer simply compares numerical values, but transforms a state point into a quantitative indicator—that is, how likely is that state point is performing compensatory regulation. Furthermore, to ensure the accuracy of the above identification under different growth stages or environmental backgrounds, this invention defines the range parameter of the membership evaluation function. The membership evaluation function is correlated with the variance of recent data. During periods of low physiological activity, the data variance and range parameter values ​​are small, making the function more sensitive to deviations from the lower bound center and able to detect subtle anomalies. Conversely, during periods of high physiological activity, the data variance and range parameter values ​​are large, thus expanding the tolerance range of the membership evaluation function. This design ensures that the membership calculation is not a rigid standard but can be adjusted based on recent activity, thereby guaranteeing the accuracy of identifying compensatory responses and avoiding misjudgments caused by changes in the crop's own physiological baseline.

[0063] Thus, the membership function of the compensatory response region was obtained by performing dynamic minimum value analysis on the preprocessed physiological state data.

[0064] Step S3: Obtain the adaptive distance adjustment factor by jointly assessing the rate of change of physiological state points and risk level.

[0065] After solving the initial identification problem of the compensatory response region in step S2, the goal of step S3 is to construct a final regulatory factor that accurately assesses the physiological differences between any two physiological state data points. To address potential misjudgments in existing solutions when dealing with complex real-world physiological processes, this invention decomposes distance regulation into two interacting but functionally different mechanisms: a risk-constrained incoordination penalty mechanism and a membership-gated homogeneity reward mechanism. The risk-constrained penalty mechanism aims to quantify the inconsistency in the dynamic trends of two state points. Plants undergo healthy physiological regulation processes to adapt to rapid environmental changes. A typical characteristic of such processes is a rapid reversal in the direction of the rate of change of their physiological indicators, thus forming an inflection point in the data. To avoid incorrectly penalizing such healthy state transitions, the penalty mechanism of this invention incorporates consideration of physiological risk levels. Its core logic is that only when a state point is already in a high-risk state should its incoordination in its trend be considered a strong negative signal and penalized; a low-risk state point in a normal regulatory process should not be penalized even if it experiences a brief reversal in direction.

[0066] The gating reward mechanism aims to address the problem of reinforcing the identification of normal physiological state points. The membership evaluation in step S2 identifies which points do not belong to the compensation zone; these points correspond to normal diurnal rhythms in the scene. This invention designs a reward function for these points, which is gated and activated by the membership evaluation value. When a pair of state points both move away from the compensation zone, the reward function is activated, actively reducing the distance between them to enhance the cohesion of normal data clusters.

[0067] In summary, firstly, a risk level analysis is performed on sucrose concentration data and abscisic acid concentration data to obtain a penalty factor. Specifically, at any target time, the rate of change of sucrose concentration at that time is obtained by subtracting the sucrose concentration values ​​at adjacent sampling points at the time following the target time from the sucrose concentration values ​​at adjacent sampling points at the time preceding the target time, and then dividing the result by the sampling time interval. Similarly, the rate of change of abscisic acid concentration at that time is obtained by subtracting the abscisic acid concentration values ​​at adjacent sampling points at the time following the target time from the abscisic acid concentration values ​​at adjacent sampling points at the time preceding the target time, and then dividing the result by the sampling time interval. The rate of change of sucrose concentration and the rate of change of abscisic acid concentration are then vectorized to obtain the rate of change vector of physiological state data at the target time. For any target time, the Mahalanobis distance between the physiological state data at the target time and the average physiological state data within the corresponding retrospective time window is used as the first risk assessment for the target time.

[0068] For any two target times, the first target time and the second target time, the average of the first risk assessment at the first target time and the first risk assessment at the second target time is used as the second risk assessment between the first and second target times; the second risk assessment is mapped using the hyperbolic tangent function, and the corresponding mapping result is used as the third risk assessment between the first and second target times; the average vector of the rate of change vectors of physiological state data at the first and second target times is used as the average rate of change vector between the first and second target times; the rate of change vector of physiological state data at the first target time... The square of the magnitude of the vector obtained by subtracting the corresponding average rate of change vector from the first target time is used as the synergistic evaluation. The square of the magnitude of the vector obtained by subtracting the rate of change vector of the physiological state data at the second target time from the corresponding average rate of change vector is used as the synergistic evaluation at the second target time. The result of adding the synergistic evaluation at the first target time and the synergistic evaluation at the second target time is used as the numerator, and the result of adding the square of the magnitude of the rate of change vector of the physiological state data at the first target time and the square of the magnitude of the rate of change vector of the physiological state data at the second target time is used as the denominator. The resulting fraction is used as the synergy evaluation between the first target time and the second target time.

[0069] The result of multiplying the third risk assessment and synergy assessment at the first and second target times is used as the penalty factor at the first and second target times.

[0070] In one implementation, assume the first The rate of change vector of physiological state data at each time point is: ;No. The rate of change vector of physiological state data at each time point is: ;No. The first risk assessment at that moment is ;No. The first risk assessment at that moment is ;No. The moment and the first The average rate of change vector at each time step is: Then the first The moment and the first The expression for calculating the penalty factor at time is:

[0071]

[0072] in, Indicates the first The moment and the first The penalty factor at each moment; Indicates the first First risk assessment at each moment; Indicates the first First risk assessment at each moment; Indicates the first The rate of change vector of physiological state data at each moment; Indicates the first A vector of the rate of change of physiological state data at each moment; Indicates the first The moment and the first The average rate of change vector at each moment; Represents the magnitude of a vector.

[0073] After obtaining the penalty factors for the first and second target times, the reward factors are obtained by performing data homogeneity analysis on the sucrose concentration data and abscisic acid concentration data. Specifically, for any two target times, the membership assessment of the physiological state data of the first target time to the compensatory response region and the membership assessment of the physiological state data of the second target time to the compensatory response region are multiplied and the square root is calculated. The corresponding calculation result is used as the joint evaluation of the membership of the first and second target times. The constant 1 is subtracted from the calculation result of the joint evaluation of the membership of the first and second target times as the gating factor for the first and second target times.

[0074] The target time with the later timestamp between the first target time and the second target time is taken as the baseline target time. The average distance between all physiological state data in the retrospective time window corresponding to the baseline target time and the mean point of all physiological state data in the retrospective time window is taken as the basic distance constant between the first target time and the second target time.

[0075] The Euclidean distance between the physiological state data corresponding to the first and second target times is used as the basic Euclidean distance between the first and second target times. The square of the basic distance constant between the first and second target times is used as the numerator, and the result of adding the square of the basic distance constant between the first and second target times to the square of the basic Euclidean distance is used as the denominator. The resulting fraction is used as the first reward assessment for the first and second target times. The result of multiplying the first risk assessment of the first target time and the first risk assessment of the second target time and calculating the square root is used as the comprehensive risk assessment for the first and second target times. The constant 1 is used as the numerator, and the result of adding the comprehensive risk assessment of the first and second target times to the constant 1 is used as the denominator. The resulting fraction is used as the second reward assessment for the first and second target times. The result of multiplying the gating factor of the first and second target times, the first reward assessment, and the second reward assessment is used as the reward factor for the first and second target times.

[0076] In one implementation, assume the first The membership assessment of physiological state data at each time point to the compensatory response region is as follows: ;No. The membership assessment of physiological state data at each time point to the compensatory response region is as follows: ;No. The moment and the first The fundamental distance constant at each time step is ;No. The moment and the first The comprehensive risk assessment at each moment is ;No. The moment and the first The fundamental Euclidean distance at each time step is Then the first The moment and the first The formula for calculating the reward factor at time point 1 is:

[0077]

[0078] in, Indicates the first The moment and the first Reward factors at each moment; Indicates the first Evaluation of the membership of physiological state data at each time point to the compensatory response region; Indicates the first Evaluation of the membership of physiological state data at each time point to the compensatory response region; Indicates the first The moment and the first The fundamental distance constant at each moment; Indicates the first The moment and the first The fundamental Euclidean distance at each moment; Indicates the first The moment and the first Comprehensive risk assessment at each moment.

[0079] Finally, by performing joint normalization on the penalty factor and the reward factor, an adaptive distance adjustment factor for differentiated distance adjustment is obtained. Specifically, for any two target times, the first target time and the second target time, the result of adding the penalty factor of the first target time and the constant 1 to the second target time is used as the numerator, and the result of adding the reward factor of the first target time and the constant 1 to the second target time is used as the denominator. The corresponding fraction is used as the adaptive distance adjustment factor of the first target time and the second target time.

[0080] In one embodiment, the first The moment and the first The formula for calculating the adaptive distance adjustment factor at time is:

[0081]

[0082] in, Indicates the first The moment and the first The penalty factor at each moment; Indicates the first The moment and the first The reward factor at each moment.

[0083] It should be noted that the molecule of the adaptive distance adjustment factor Responsible for imposing penalties, denominator This factor is responsible for applying rewards. Through the synergistic effect of punishment and reward factors, it differentiates between different types of physiological state pairs. First, the punishment factor aims to address the problem of accurately identifying true stress states while avoiding misjudgments of healthy physiological regulatory processes. This function consists of a risk gating term. Inco-measures with normalization The composition is as follows: For point pairs whose physiological rhythms are disrupted due to real stress, the risk level is significantly increased, while the coherence of their rate of change is poor. These two conditions cause the penalty factor to calculate a significantly positive value greater than 0, thus affecting the final moderating factor. It is much greater than 1. This achieves effective punishment of the real stress point pairs, significantly amplifying their distance. For healthy points in low-risk states, even if their change trends are temporarily uncoordinated due to changes in response to environmental changes (such as the inflection point of the adjustment process), their extremely low risk level will cause the risk gating term to approach 0, thereby causing the value of the punishment factor to approach 0, avoiding incorrect punishment for such healthy states.

[0084] Secondly, the reward factor is a targeted similarity attraction mechanism designed to address the problem of reinforcing the recognition of normal physiological states. Its activation is controlled by gating terms. To control. When a pair of state points are both far from the compensatory response region, it is highly likely that they belong to a normal diurnal variation rhythm ( When the value approaches 0, the reward factor is activated. At this point, the reward factor, through an inverse distance ratio, affects the state value as it approaches ( Small and low risk level For normal point pairs (small), a larger positive reward value is calculated. This allows the adjustment factor to... denominator Significantly larger than the molecule (at this time ), ultimately leading to The factor is much less than 1. This effectively reduces the distance between normal state points, enhances the cohesion of normal data clusters, and helps to more clearly separate outliers during subsequent clustering. Finally, the design of this invention can also correctly handle healthy compensatory response points. Since the physiological change trends of these points are coordinated, their non-coordination measure will be very small, causing the penalty factor to approach 0. At the same time, since these points are located within the compensation region ( If the term approaches 1, the gating activator of its reward factor will approach 0, causing the reward factor to also approach 0. Therefore, the final adjustment factor... The value will approach 1. This means that the design of the present invention neither incorrectly punishes healthy compensatory response points nor incorrectly imposes rewards on them as would be typically given to normal rhythmic points, but rather neutrally maintains their original distance relationships.

[0085] Thus, the adaptive distance adjustment factor was obtained by jointly assessing the rate of change of physiological state points and risk levels.

[0086] Step S4: By introducing an adaptive distance adjustment factor into the Euclidean distance calculation, the optimized distance metric is obtained and cluster diagnosis is performed.

[0087] This step applies the adaptive distance adjustment factor obtained in step S3 to the distance calculation between physiological state data. For any two target times, the first target time and the second target time, the square root of the product of the square of the basic Euclidean distance between the first target time and the second target time and the adaptive distance adjustment factor is used as the optimized distance metric between the first target time and the second target time. Cluster analysis is then performed using the optimized distance metric between the first target time and the second target time to obtain cluster diagnosis results. It should be noted that in this embodiment of the invention, cluster analysis is performed using the DBSCAN clustering algorithm, and the corresponding clustering results are used as cluster diagnosis results. The output results divide all physiological state data points into one or more high-density core clusters (representing normal physiological states) and several outliers (representing abnormal or physiological states requiring special attention).

[0088] This completes the process of introducing an adaptive distance adjustment factor into Euclidean distance calculation, obtaining an optimized distance metric, and performing cluster diagnosis.

[0089] Step S5: By performing state mapping on the cluster diagnosis results, obtain the photosynthetic dynamic prediction conclusions and regulation strategies.

[0090] By directly mapping the diagnostic conclusions output in step S4 to qualitative predictions of future photosynthetic trends and corresponding regulatory strategy suggestions, for any target time to be analyzed, the diagnostic label of its state point is first determined based on the clustering results of step S4.

[0091] Based on this diagnostic label, the system can output a prediction of the photosynthetic dynamics trend in the near future. If the diagnostic label at the target time is "normal state," the predicted photosynthetic dynamics will mainly follow the regular diurnal variation pattern determined by current environmental conditions such as light intensity. If the diagnostic label at the target time is "outlier," and further confirmation is made based on the membership assessment in step S2 to determine if it is in a "compensatory regulation state," then it is predicted that it will maintain a relatively high stomatal opening and photosynthetic efficiency in the upcoming low-light environment, exhibiting a trend of efficient light energy utilization. If the target time is confirmed as a "potential stress state," then its future photosynthetic activity will be inhibited, and even if environmental conditions are suitable, its photosynthetic rate will be difficult to reach the optimal level. These predictive conclusions can be directly used as a basis for decision-making, providing clear regulatory guidance for the automated control system. For corn in a "compensatory regulation state," the system should avoid unnecessary stress warnings or interventions; while for corn in a "potential stress state," stress checks and mitigation procedures should be triggered first, rather than blindly performing yield-inducing operations such as supplemental lighting.

[0092] Thus, the process of obtaining photosynthetic dynamic prediction conclusions and regulation strategies by performing state mapping on cluster diagnostic results has been completed.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting photosynthetic dynamic regulation based on temporal changes in sap concentration, characterized in that, The method for predicting photosynthetic dynamic regulation based on temporal changes in sap concentration includes: Step S1: Obtain smooth time-series physiological state data by preprocessing the sucrose and abscisic acid concentration data of corn sap. Step S2: Obtain the membership function of the compensatory response region by performing dynamic minimum analysis on the preprocessed physiological state data; Step S3: Obtain the adaptive distance adjustment factor by jointly assessing the rate of change of physiological state points and risk level; Step S4: By introducing an adaptive distance adjustment factor into the Euclidean distance calculation, the optimized distance metric is obtained and cluster diagnosis is performed; Step S5: By performing state mapping on the cluster diagnosis results, obtain the dynamic prediction conclusions and regulation strategies for photosynthesis.

2. The photosynthetic dynamic regulation prediction method based on the temporal variation of sap concentration according to claim 1, characterized in that, The process involves preprocessing data on sucrose and abscisic acid concentrations in corn sap to obtain smooth, time-series physiological state data, including: Set the sensor sampling frequency; deploy sap flow sensors in the corn growing environment and continuously collect time-series data on sucrose and abscisic acid concentrations in corn stalk sap according to the sensor sampling frequency; deploy an environmental sensor network in the corn growing environment to synchronously collect environmental parameter data, including light intensity data, air temperature data, and air humidity data; input the collected raw time-series data into a preprocessing process, which includes: using... The criteria involve scanning the time-series data of sucrose concentration and abscisic acid concentration, identifying and removing outliers; filling in missing data points after outlier removal using linear interpolation; and smoothing the corrected time-series data of sucrose concentration and abscisic acid concentration using moving average filtering to obtain preprocessed time-series data of sucrose concentration and abscisic acid concentration.

3. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 1, characterized in that, The step of obtaining the membership function of the compensatory response region by performing dynamic minimum value analysis on the preprocessed physiological state data includes: By setting a retrospective time window length and performing retrospective time window analysis on sucrose concentration data and abscisic acid concentration data, the characteristic center value of the compensatory response region is obtained. For any target time point in the sucrose concentration time series data and abscisic acid concentration time series data, the membership assessment of the physiological state data at the target time point to the compensatory response region is obtained by performing variance evaluation and normalization on the sucrose concentration data and abscisic acid concentration data at the target time point based on the characteristic central value of the compensatory response region.

4. The photosynthetic dynamic regulation prediction method based on the temporal variation of sap concentration according to claim 3, characterized in that, The process involves retrospectively analyzing sucrose concentration data and abscisic acid concentration data over a time window to obtain the characteristic center values ​​of the compensatory response region, including: For any target time in the time-series physiological state data, preprocessed sucrose concentration data from consecutive sampling times are selected within the retrospective time window corresponding to the target time. The minimum value of the sucrose concentration data within this time window is obtained, and this minimum value is used as the feature center value of the compensatory response region in the sucrose concentration dimension. Preprocessed abscisic acid concentration data from consecutive sampling times are selected within the target time window. The minimum value of the abscisic acid concentration data within this time window is obtained, and this minimum value is used as the feature center value of the compensatory response region in the abscisic acid concentration dimension.

5. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 3, characterized in that, The process involves evaluating and normalizing the sucrose and abscisic acid concentration data at the target time based on the characteristic center values ​​of the compensatory response region to obtain the membership assessment of the physiological state data at the target time to the compensatory response region, including: The variance of sucrose concentration data in the retrospective time window corresponding to the target time is used as the range parameter for the sucrose concentration dimension; the variance of abscisic acid concentration data in the retrospective time window corresponding to the target time is used as the range parameter for the abscisic acid concentration dimension. The numerator is the result of subtracting the feature center value of the compensatory response region at the target time from the sucrose concentration data at the target time in the sucrose concentration dimension. The denominator is the square of the range parameter of the sucrose concentration dimension at the target time. The resulting fraction is used as the first evaluation factor of the sucrose concentration at the target time. The numerator is the result of subtracting the feature center value of the compensatory response region in the abscisic acid concentration dimension from the abscisic acid concentration data at the target time, and the denominator is the square of the range parameter of the abscisic acid concentration dimension at the target time. The resulting fraction is used as the first evaluation factor of the abscisic acid concentration at the target time. The negative of the result of subtracting the first evaluation factor of the sucrose concentration at the target time from the first evaluation factor of the abscisic acid concentration at the target time is used for exponential mapping with the natural constant as the base. The corresponding mapping result is used as the membership assessment of the physiological state data at the target time to the compensatory response region.

6. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 1, characterized in that, The process of obtaining an adaptive distance adjustment factor by jointly assessing the rate of change of physiological state data and risk level includes: By conducting risk level analysis on sucrose concentration data and abscisic acid concentration data, penalty factors were obtained. Reward factors were obtained by performing data homogeneity analysis on sucrose concentration data and abscisic acid concentration data; By jointly normalizing the penalty factor and the reward factor, an adaptive distance adjustment factor for differentiated distance adjustment is obtained.

7. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 6, characterized in that, The step involves analyzing the risk level of sucrose concentration data and abscisic acid concentration data to obtain penalty factors, including: At any target time, the rate of change of sucrose concentration at the target time is obtained by subtracting the sucrose concentration at the adjacent sampling point at the next time from the sucrose concentration at the previous time from the sucrose concentration at the next time, and dividing the result by the sampling time interval. The rate of change of abscisic acid concentration at the target time is obtained by subtracting the abscisic acid concentration at the adjacent sampling point at the next time from the abscisic acid concentration at the previous time from the abscisic acid concentration at the next time. The rate of change of sucrose concentration at the target time and the rate of change of abscisic acid concentration at the target time are vector-synthesized to obtain the rate of change vector of physiological state data at the target time. For any target time, the Mahalanobis distance between the physiological state data at the target time and the average physiological state data in the corresponding retrospective time window is used as the first risk assessment for the target time; for any two target times, the first target time and the second target time, the mean of the first risk assessment at the first target time and the first risk assessment at the second target time is used as the second risk assessment between the first target time and the second target time; the second risk assessment is mapped using the hyperbolic tangent function, and the corresponding mapping result is used as the third risk assessment between the first target time and the second target time; The average vector of the rate of change vectors of physiological state data at the first target time and the rate of change vectors of physiological state data at the second target time is used as the average rate of change vector at the first target time and the second target time. The square of the magnitude of the vector obtained by subtracting the rate of change vector of physiological state data at the first target time from the corresponding average rate of change vector is used as the co-evaluation at the first target time. The square of the magnitude of the vector obtained by subtracting the rate of change vector of physiological state data at the second target time from the corresponding average rate of change vector is used as the co-evaluation at the second target time. The numerator is the result of adding the collaborative evaluation at the first target time and the collaborative evaluation at the second target time. The denominator is the result of adding the square of the magnitude of the rate of change vector of the physiological state data at the first target time and the square of the magnitude of the rate of change vector of the physiological state data at the second target time. The corresponding fraction is used as the collaborative evaluation between the first target time and the second target time. The result of multiplying the third risk assessment and synergy assessment at the first and second target times is used as the penalty factor at the first and second target times.

8. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 6, characterized in that, The reward factor is obtained by performing data homogeneity analysis on sucrose concentration data and abscisic acid concentration data, including: For any two target times, the first target time and the second target time, the membership evaluation of the physiological state data of the first target time to the compensatory response region and the membership evaluation of the physiological state data of the second target time to the compensatory response region are multiplied and the square root is calculated. The corresponding calculation result is used as the joint evaluation of the membership of the first target time and the second target time. Subtracting the calculated result of the joint evaluation of the membership degrees of the first and second target times from the constant 1 is used as the gating factor for the first and second target times. The target time with the later timestamp between the first target time and the second target time is taken as the baseline target time. The average distance between all physiological state data in the retrospective time window corresponding to the baseline target time and the mean point of all physiological state data in the retrospective time window is taken as the basic distance constant between the first target time and the second target time. The Euclidean distance between the physiological state data corresponding to the first target time and the second target time is used as the basic Euclidean distance between the first target time and the second target time. The square of the basic distance constant between the first target time and the second target time is used as the numerator, and the sum of the square of the basic distance constant between the first target time and the second target time and the square of the basic Euclidean distance is used as the denominator. The corresponding fraction is used as the first reward evaluation for the first target time and the second target time. The result of multiplying the first risk assessment at the first target time and the first risk assessment at the second target time and taking the square root is used as the comprehensive risk assessment at the first target time and the second target time; the result of adding the comprehensive risk assessment at the first target time and the second target time to the constant 1 as the numerator and the constant 1 as the denominator is used as the corresponding fraction as the second reward assessment at the first target time and the second target time. The calculation result of multiplying the gating factors of the first target time and the second target time, and the first reward evaluation and the second reward evaluation, is used as the reward factors of the first target time and the second target time.

9. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 6, characterized in that, The process of obtaining an adaptive distance adjustment factor for differentiated distance adjustment by jointly normalizing the rate of change vector and the risk level index includes: For any two target times, the first target time and the second target time, the result of adding the penalty factor of the first target time and the constant 1 to the second target time is used as the numerator, and the result of adding the reward factor of the first target time and the constant 1 to the second target time is used as the denominator. The corresponding fraction is used as the adaptive distance adjustment factor of the first target time and the second target time.

10. The method for predicting photosynthetic dynamic regulation based on the temporal variation of sap concentration according to claim 1, characterized in that, The process of introducing an adaptive distance adjustment factor into Euclidean distance calculation to obtain an optimized distance metric and performing cluster diagnosis includes: For any two target times, the first target time and the second target time, the square root of the product of the square of the basic Euclidean distance between the first target time and the second target time and the adaptive distance adjustment factor is used as the optimized distance metric between the first target time and the second target time. Cluster analysis is performed using the optimized distance metric between the first target time and the second target time to obtain the cluster diagnosis results.

Images