Method for rapidly detecting and dynamically evaluating honey secreting amount of nectariferous plant by using near infrared spectroscopic analysis

Abstract

The invention relates to the technical field of detection and evaluation, in particular to a method for rapidly detecting and dynamically evaluating the honey secreting amount of a nectariferous plant by utilizing near infrared spectroscopic analysis, which comprises the following steps of: determining a flowering stage of a target nectariferous plant flower; collecting spectral data of the target nectariferous plant flowers in the current flowering phase by using a near infrared spectrometer; inputting the spectral data into a stage specific honey secretion amount analysis model corresponding to the current flowering stage, and outputting a honey secretion amount predicted value of the current flowering stage; in a target flowering phase, obtaining a time sequence formed by a plurality of honey secretion amount predicted values; and generating a flowering period honey secretion dynamic curve of the target honey source plant flowers according to the time sequence. According to the method, deep coupling of the model structure and the plant physiological rhythm is achieved, the prediction result is made to better conform to the actual honey secretion mechanism, higher generalization ability is achieved in different stages, and particularly, the problem of misjudgment or sudden precision drop of a traditional model can be effectively solved in the non-peak period.

Description

Technical Field

[0001] This invention relates to the field of detection and evaluation technology, and in particular to a method for rapid detection and dynamic evaluation of nectar secretion in nectar-producing plants using near-infrared spectroscopy. Background Technology

[0002] Nectar plants are the fundamental resources for the sustainable development of the bee industry, and their nectar secretion capacity directly determines the bee's foraging efficiency and honey production. To scientifically plan beekeeping areas, select high-quality nectar varieties, and rationally schedule honey harvesting periods, it is urgent to accurately assess the nectar secretion capacity of different nectar plants. Existing methods for detecting nectar secretion mainly rely on manual weighing, saccharimeter measurement, or indirect estimation methods. These methods typically suffer from significant drawbacks, including complex operation, low efficiency, susceptibility to environmental interference, and difficulty in achieving continuous dynamic observation, making them particularly difficult to apply in large-scale field environments.

[0003] On the other hand, the nectar secretion behavior of nectar-producing plants has obvious stage-specific characteristics during the flowering period. That is, there are significant differences in their physiological metabolic state, nectar gland secretion level and environmental response mechanism during the initiation, peak and decline of nectar secretion. Therefore, the relationship between nectar secretion and plant tissue structure or biochemical components exhibits non-linear and stage-specific coupling characteristics. Traditional unified modeling or static evaluation methods are difficult to effectively capture and characterize this dynamic change pattern.

[0004] In recent years, near-infrared spectroscopy has been increasingly applied to the determination of indicators such as water, sugar, and protein in plant tissues due to its advantages of being rapid, non-destructive, and applicable in the field. However, in the assessment of nectar secretion in nectar-producing plants, existing technologies mostly focus on constructing unified models for single time points and the entire life cycle, lacking targeted modeling methods that combine with the stage-specific characteristics of plant flowering. This results in limited model prediction accuracy and weak versatility, making it difficult to meet the needs for continuous dynamic monitoring and high-precision assessment of the entire nectar secretion process. Summary of the Invention

[0005] This invention provides a method for rapid detection and dynamic evaluation of nectar secretion in nectar-producing plants using near-infrared spectroscopy. By combining the physiological rhythm of plant nectar secretion with the capabilities of near-infrared spectroscopy, this method can rapidly predict nectar secretion capacity at different flowering stages and assess dynamic potential throughout the entire process, thereby improving the scientific rigor, accuracy, and application of nectar source quality evaluation.

[0006] A rapid detection and dynamic assessment method for nectar secretion from nectar-producing plants using near-infrared spectroscopy includes the following steps: S1. Flowering Stage Identification and Near-Infrared Spectroscopy Acquisition: Determine the flowering stage of the target nectar source plant flowers; the flowering stage includes at least the nectar secretion initiation stage, the nectar secretion peak stage, and the nectar secretion decline stage; use a near-infrared spectrometer to acquire spectral data of the target nectar source plant flowers at the current flowering stage; S2. Generation of stage-specific nectar secretion prediction values: The spectral data is input into the stage-specific nectar secretion analysis model corresponding to the current flowering stage, and the predicted nectar secretion value for the current flowering stage is output; the stage-specific nectar secretion analysis model is a prediction model independently established for different flowering stages; S3. Generation and potential assessment of nectar secretion dynamic curve: During the target flowering period, S1 to S2 are repeated at a predetermined frequency to obtain a time series consisting of multiple predicted nectar secretion values; a nectar secretion dynamic curve of the target nectar source plant flowers is generated based on the time series; based on the morphological characteristics of the nectar secretion dynamic curve, its nectar secretion pattern and total nectar secretion potential are assessed.

[0007] Optionally, determining the flowering stage includes acquiring a visible light image of the target flower, analyzing its corolla opening, petal color and morphological characteristics through an image recognition unit, and matching it with a database of flowering stage morphological characteristics of the target nectar source plant flower varieties to determine its flowering stage.

[0008] Optionally, the step of using a near-infrared spectrometer to collect spectral data of the target nectar source plant flowers during the current flowering stage includes: Align the probe of the near-infrared spectrometer vertically with and lightly touch the nectary area or the base of the stamen to ensure that the position is consistent for each measurement. During the current flowering period, at fixed times each day, under conditions of avoiding direct sunlight, multiple spectral scans are performed on the same flower or the same batch of sample flowers, and the results of the multiple scans are averaged to obtain a spectral data representing the current flowering period.

[0009] Optionally, the stage-specific nectar secretion analysis model is constructed using a partial least squares regression algorithm. It utilizes the correlation between nectar secretion and spectral data in the current flowering stage sample to extract the spectral principal components related to the response variable, and establishes a linear regression model based on the spectral principal components.

[0010] Optionally, S2 further includes preprocessing the spectral data by performing a standard normal transformation, wherein the standard normal transformation is performed by centering the mean and normalizing the standard deviation of the spectral vector of each sample, and then inputting the processed data into the model.

[0011] Optionally, S2 specifically includes: The preprocessed spectral data, based on the flowering stage identified in S1, is input into a stage-specific nectar secretion analysis model corresponding to the current flowering stage. The model outputs a quantitative nectar secretion prediction value, which characterizes the instantaneous nectar secretion capacity of the flower unit at the current flowering stage. The stage-specific nectar secretion analysis model is established independently for each different flowering stage as follows: Flower samples known to be in the nectar secretion initiation, peak nectar secretion, and nectar secretion decline stages were collected as the training set; Spectral data of each training set sample was collected using a near-infrared spectrometer, and its actual nectar secretion was simultaneously measured using the standard weight method to form a spectral-nectar secretion calibration dataset. For the calibration datasets of the initiation period, peak period, and decline period of nectar secretion, the partial least squares regression algorithm was used to train three distinct and optimized quantitative calibration models, namely the stage-specific nectar secretion analysis models.

[0012] Optionally, the stage-specific nectar secretion analysis model may also include the introduction of a sparse regularization term to compress redundant band features during the regression coefficient solution process, retaining only the key spectral variables that contribute to the prediction of nectar secretion.

[0013] Optionally, in S3, with a fixed 24-hour cycle, S1 to S2 are repeated on the same batch of pre-marked target nectar plant flowers at the same fixed time each day until the end of the flowering period. The predicted nectar secretion value output after each execution is associated with the corresponding collection date and timestamp, and stored in chronological order to form a time-series sequence of predicted nectar secretion values ​​ordered by time.

[0014] Optionally, generating the dynamic curve of nectar secretion during the flowering period of the nectar-producing plant based on the time sequence specifically includes: In a coordinate system with time as the horizontal axis and the predicted nectar secretion value as the vertical axis, each data point in the time series is labeled; a curve fitting algorithm is used to connect all data points to generate a smooth and continuous dynamic change curve of nectar secretion during the flowering period, so as to intuitively show the evolution trend of nectar secretion with the flowering period.

[0015] Optionally, the assessment of the nectar secretion pattern and total nectar secretion potential based on the morphological characteristics of the nectar secretion dynamic curve during the flowering period specifically includes: Nectar secretion pattern assessment: Analyze the morphological characteristics of the curve, including identifying the peak point of the curve to determine the occurrence time of the peak secretion period, calculating the duration for which the peak of the curve exceeds the set high nectar secretion threshold to assess the duration of efficient nectar secretion, and analyzing the slope of the rising and falling branches of the curve to determine the rate of nectar secretion initiation and the speed of decline. Total nectar secretion potential assessment: The area of ​​the closed region enclosed by the dynamic curve of nectar secretion during flowering and the horizontal axis is calculated by numerical integration method, and the integrated area is used as a quantitative indicator to evaluate the total nectar secretion potential of the nectar source plant.

[0016] The beneficial effects of this invention are: This invention employs near-infrared spectroscopy for non-contact, rapid detection of nectar-producing plant flowers. Combined with preprocessing algorithms and stage-specific modeling methods, it can predict the nectar secretion per unit flower in real time without damaging the flower structure or performing traditional sampling measurements, thus achieving high-throughput, on-site, and automated assessment of nectar secretion capacity.

[0017] This invention addresses the problem that traditional unified models for the entire nectar secretion cycle struggle to accurately capture significant spectral differences and inconsistent data distributions across different nectar secretion stages. It proposes a stage-specific nectar secretion analysis model structure, constructing independent spectral regression models for the nectar secretion initiation, peak, and decline stages. This fully leverages highly correlated variables and principal component characteristics within each stage, achieving deep coupling between the model structure and plant physiological rhythms. This makes the prediction results more consistent with the actual nectar secretion mechanism and exhibits stronger generalization ability across different stages. Particularly during non-peak periods, it effectively avoids the misjudgment or sharp drop in accuracy issues common in traditional models.

[0018] This invention, based on single-point nectar secretion prediction, constructs a complete dynamic change curve of nectar secretion during the flowering period through timed sequence sampling and curve fitting. Furthermore, it introduces various physiologically structured characteristic indicators such as the duration of high nectar secretion, peak time point, nectar secretion rate slope, and total integral. This not only visualizes the nectar secretion rhythm of nectar-producing plants but also provides a key means to scientifically quantify nectar secretion potential, overcoming the limitations of previous evaluations relying solely on single-point or average nectar secretion amounts. It provides strong technical support for applications such as beekeeping plan optimization, nectar source selection decisions, and plant breeding and selection. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the method execution logic in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0022] like Figures 1-2As shown, a rapid detection and dynamic evaluation method for nectar secretion in nectar-producing plants using near-infrared spectroscopy includes the following steps: S1. Flowering stage identification and near-infrared spectral acquisition: Determine the flowering stage of the target nectar source plant flowers; the flowering stage includes at least the nectar secretion initiation stage, the nectar secretion peak stage, and the nectar secretion decline stage; use a near-infrared spectrometer to acquire spectral data of the target nectar source plant flowers at the current flowering stage.

[0023] S11, flower stage identification: Determine the flowering stage of the target nectar source plant, including one or more of the following discrimination methods.

[0024] Method 1: Based on biologically effective accumulated temperature: Obtain the cumulative biologically effective accumulated temperature of the target flower. And compared with the accumulated temperature threshold range of each flowering stage of the nectar source plant. By comparing the results, we can determine the flowering stage of the plant.

[0025] The expression is as follows: ; in, Indicates biologically effective accumulated temperature. , Indicates the first The day's highest and lowest temperatures, The reference temperature (i.e., the physiological zero degree for this plant variety) The cumulative number of days from the budding date, and the accumulated temperature threshold range during the flowering period, if satisfying the following: Then determine that the target flower is in the first position. Flowering stages (initiation, peak, or decline of nectar secretion).

[0026] Method 2: Visible light image-based recognition and discrimination: Acquire RGB images of the target flower. The data is input into the image recognition unit to extract its morphological feature parameter set. ,include: Corolla opening: Petal color RGB value or HSV hue; Texture indicators such as the serration, curvature, and pointed / rounded shape of the petal edges; The above feature vectors are compared with a pre-established flowering stage-morphological feature database. Comparisons were performed, and the flowering period was identified using either the least Euclidean distance or a classification model. ;in, Flowering stages in the database The standard morphological feature vector.

[0027] Specifically: First, using a visible light camera or RGB industrial camera with high-definition imaging capabilities, images of typical flowers of the target nectar source plant are acquired to obtain clear, frontal images. Image acquisition should be carried out under soft natural light and a clean background to ensure that the flower's structural outline is complete and discernible. After acquisition, the images are input into the image recognition unit of the built-in image analysis model.

[0028] The image recognition unit performs flower region segmentation and feature extraction operations on the input image. The extracted morphological features of the flower include the following categories: Corolla opening angle: The opening angle of the corolla is calculated by edge detection and angle fitting algorithm to determine whether the flower is fully open; Petal color and hue: Extract the dominant color value of the petal area and convert it to the hue channel in the HSV color space to identify whether the petal has entered the fading stage; Petal texture and morphological characteristics: Analyze indicators such as the smoothness, serration, and rounded / sharp shape of the petal edges to identify whether it is in the blooming or withering stage; Flower stamen exposure level: Determine whether the flower stamen is fully exposed, which helps to determine the active period of nectar secretion.

[0029] The extracted feature values ​​are used to form a set of morphological feature vectors. These vectors are compared with a pre-established flowering stage-morphological feature database, which stores the mean sample of typical image features of the nectar source plant at different flowering stages (nectar secretion initiation stage, nectar secretion peak stage, and nectar secretion decline stage).

[0030] The comparison method can use the nearest neighbor Euclidean distance calculation method for similarity matching, and finally output the flowering period discrimination result, that is, determine which stage the flower in the current image belongs to, so as to provide stage label support for subsequent near-infrared spectral acquisition and nectar secretion prediction.

[0031] The establishment of the flowering stage-morphological characteristic database is as follows: Throughout the flowering period of nectar-producing plants, representative healthy flowers were systematically sampled at different stages, including the initiation, peak, and decline of nectar secretion. A large number of representative frontal images of flowers were collected at each stage, ensuring image clarity, uniform angle, and suitable lighting. Image analysis algorithms were used to extract a set of stable floral morphological feature parameters from each image, such as corolla opening angle, petal color, petal edge contour features, texture clarity, and stamen exposure. The extraction results for each image were recorded as a feature vector. Statistical analysis was performed on the large number of feature vectors extracted at each flowering stage, calculating the mean, standard deviation, and other statistical indicators for each stage, and saving them as standard morphological templates in the database. The final database uses flowering stage as the classification label, with each category corresponding to a set of morphological feature statistical templates for stage matching reference in subsequent actual image recognition. This database can be regularly expanded and updated according to variety, growing region, and climatic conditions.

[0032] S12, Near-infrared spectral acquisition during the current flowering stage: The near-infrared spectrometer probe is vertically aligned with the nectar-related structural areas of the target flower, such as the nectary region or the base of the stamens, ensuring slight contact and fixation of the probe to obtain a stable and consistent sampling signal. Within the currently identified flowering stage, spectral scanning is repeatedly performed on the same flower or a batch of representative sample flowers at fixed times each day, avoiding direct sunlight. Next, obtain multiple raw spectra: ; The representative spectrum for the current flowering period is obtained by averaging the multiple spectral scans along the wavelength dimension. ; final This refers to the characteristic spectral data of the input stage-specific nectar secretion analysis model.

[0033] S2. Generation of stage-specific nectar secretion prediction value: The spectral data is input into the stage-specific nectar secretion analysis model corresponding to the current flowering stage, and the predicted nectar secretion value for the current flowering stage is output; the stage-specific nectar secretion analysis model is a prediction model independently established for different flowering stages.

[0034] S2 mainly involves inputting spectral data into a stage-specific nectar secretion analysis model corresponding to the current flowering stage, and outputting predicted nectar secretion values ​​as follows: First, based on the flowering stage identified through accumulated temperature analysis, image recognition, or spectral absorption peak ratio, the current physiological state of the flower is determined, such as the nectar secretion initiation stage, peak nectar secretion stage, or nectar secretion decline stage. Based on this determination, a dedicated nectar secretion analysis model pre-established for this specific flowering stage is automatically invoked. Each model is independently trained for that stage and is specifically designed to depict the precise correspondence between plant nectar secretion and spectral response during that period, avoiding the problem of large prediction errors in traditional full-cycle unified models during off-peak periods.

[0035] Then, the spectral data after standard preprocessing is input into the model. First, the spectral principal components most correlated with the amount of nectar secretion are extracted. This process is achieved using the partial least squares regression algorithm, which can screen out the band combination that best represents the changes in nectar secretion in the high-dimensional spectral space. Subsequently, the principal component variables are input into the regression structure of the model to calculate the predicted value of nectar secretion, which represents the amount of nectar that the flower may secrete per unit time.

[0036] In this process, each model has an independent feature compression structure and regression weight coefficients. Sparse constraints are introduced during the training of model parameters to effectively enhance the model's ability to suppress abnormal bands and its sensitivity to key bands, thereby improving the model's predictive stability and generalization ability under different flowering periods, different individuals, and microenvironmental changes.

[0037] This invention employs a phased modeling approach, breaking away from the limitations of previous single models that covered the entire life cycle. By deeply coupling the dynamic physiological state of plants with their spectral responses, the model not only possesses high-precision predictive capabilities but also exhibits excellent biological interpretability and spectral engineering adaptability, thereby enhancing the scientific rigor and application value of nectar secretion assessment in nectar-producing plants.

[0038] S2 includes the following sub-steps: S21, the preprocessed spectral data The input into the analysis model corresponding to the current flowering stage includes the following steps: Based on the current flowering stage determined in S1 Start-up phase, peak phase, decline phase Select and call the corresponding stage-specific model ; spectral feature vector Input the data into the selected model, and the output will be the predicted nectar secretion rate for the current flowering period. : Where S represents the representative average spectral data vector of the current flowering stage, with dimension 1. (Number of spectral channels) This represents the specific partial least squares regression (PLSR) model corresponding to the current flowering stage. This represents the predicted nectar secretion value output by the model, indicating the instantaneous nectar secretion capacity of a single flower at the current stage.

[0039] S22, To ensure the accuracy of the prediction, the stage-specific nectar secretion analysis model is constructed independently mainly through the following methods: S221, Sample Collection and Label Construction: A large number of flower-bearing samples were collected during the three typical nectar secretion stages of nectar-producing plants (initiation, peak, and decline). Record its flowering period label And perform the following for each sample: Spectral data of its flowers were collected using a near-infrared spectrometer. ; The true nectar secretion of the sample was simultaneously obtained using either the standard gravimetric method or the saccharimetric method. ; S222, Data calibration set construction: For each stage Start-up phase, peak phase, decline phase Construct the following dataset: ; Indicates the first The near-infrared spectral vector (reflectance or absorbance at a certain wavelength) of a sample flower. Indicates the first The actual nectar secretion of each sample flower; S223, Model Training (Partial Least Squares Regression): For each stage dataset, the regression model is trained using the PLSR algorithm. PLSR establishes a linear mapping relationship by extracting the most relevant latent variables (principal components) between spectral data and target nectar secretion.

[0040] Compared to the traditional uniform model for the entire flowering period This method employs a phased modeling strategy, namely: Average Fit Large error fluctuations; Accurately fit a specific stage The error was significantly reduced; Physiological metabolic activities caused by changes in flowering period have different spectral response characteristics at different stages. This method allows for the separate extraction of principal components and variable sets for each stage, avoiding "model compromise" caused by excessive differences in data distribution between stages.

[0041] More specifically: The model structure consists of building an independent nectar secretion prediction model for each flowering stage (nectar secretion initiation, peak, and decline). The overall form of the trained model is as follows: ;in, This represents the principal component extraction transformation performed on the input spectral data. This represents the regression coefficient vector of the stage-specific model. Represents the model bias term. This represents the PLSR nectar secretion prediction function corresponding to the current flowering stage.

[0042] The training process for each model is strictly limited to its corresponding flowering stage. The specific data construction process is as follows: 1. Experimental Data Collection: Sample selection: Collect a large number of representative flower samples at specific flowering stages under natural conditions and record their labels (nectar secretion stage identifiers).

[0043] Spectral measurement: The nectary region of the flower was scanned multiple times using a handheld near-infrared spectrometer (wavelength range 700–1100 nm, resolution ≤5 nm) to form standard spectral data on average.

[0044] Actual measurement of honey secretion: The honey secreted per unit time was collected using a micro balance and weighed accurately; Get each sample pair .

[0045] 2. Spectral data preprocessing: Background and baseline drift are eliminated using a first-order derivative (SG filter), and a standard normal transformation (SNV) is performed on each spectrum. ; This represents the mean of the spectral vector of the sample (i.e., the average value at all wavelengths). This represents the standard deviation of the spectral vector of the sample. (For a specific stage dataset) Perform partial least squares regression (PLSR) modeling separately to extract response variables. Highly correlated principal components (first 5-10): .

[0046] The model training process is as follows: 1. Partial Least Squares Regression (PLSR) uses the PLSR method to find the most correlated latent variables when constructing a regression model, reducing the original high-dimensional spectral data to principal component matrices. And perform linear regression: ; in, This indicates the loading matrix (principal component direction). Represents the score matrix (sample principal components). Represents the regression coefficient. The final model for the bias term is: ; 2. To further improve the model's interpretability and spectral variable selectivity, PLSR can be replaced with sparse partial least squares regression (sPLSR), and the objective function can be introduced... Regularization: ;in, The sparsity regularization strength is determined through cross-validation. The L1 norm constraint makes most of the weights zero, retaining only the key bands.

[0047] S3. Generation and potential assessment of nectar secretion dynamic curve: During the target flowering period, S1 to S2 are repeated at a predetermined frequency to obtain a time series consisting of multiple predicted nectar secretion values; a nectar secretion dynamic curve of the target nectar source plant flowers is generated based on the time series; based on the morphological characteristics of the nectar secretion dynamic curve, its nectar secretion pattern and total nectar secretion potential are assessed.

[0048] S31, Generation of time-series nectar secretion prediction value sequence: During the target flowering period, with fixed daily time periods as observation nodes and a 24-hour sampling cycle, S1 to S2 were repeatedly executed on the same batch of marked target flowers until the end of the flowering period. After each execution, the current predicted nectar secretion value was linked to the collection time to construct a time-series data pair. ;in, Indicates the first The timestamp of the last collection. This represents the predicted nectar secretion rate (mg / flower / hour) at the corresponding time point. This indicates the total number of sampling operations performed during this flowering period. This represents a time series of predicted nectar secretion values ​​sorted by time.

[0049] S32, Generation of dynamic curve of nectar secretion during flowering: Using time as the horizontal axis The predicted nectar secretion value is on the vertical axis. The above time series data points are plotted in a two-dimensional coordinate system, and the points are connected by a fitting function to generate a nectar secretion trend curve during the flowering period.

[0050] Fitting methods include spline interpolation: ;in, This represents the fitted dynamic function of nectar secretion. Represents any point in time The curve visually illustrates the entire dynamic process of nectar secretion, from its initiation, rise, peak, decline, and termination.

[0051] S32 aims to construct a visualized dynamic curve from the predicted nectar secretion values ​​collected daily during the flowering period, revealing the temporal evolution of nectar secretion in nectar-producing plants. Data points collected at consecutive time points are plotted on a two-dimensional coordinate system with time on the horizontal axis and the predicted nectar secretion value on the vertical axis. Subsequently, a fitting algorithm is used to smoothly connect these discrete data points; common methods include cubic spline interpolation, thereby generating a continuous and smooth dynamic curve.

[0052] This curve not only vividly reflects the entire process of nectar-producing plants from the start of nectar secretion, rapid growth, reaching the peak of nectar secretion, to gradual decline and termination, but also, through morphological characteristic analysis, can further assist in judging the nectar secretion potential, nectar secretion rhythm, and the optimal time period of the nectar secretion window, providing decision support for apiary honey harvesting arrangements and nectar plant resource evaluation.

[0053] S33, Assessment of Nectar Secretion Patterns: Based on the fitted dynamic curve f(t), the following feature parameters are extracted to evaluate the nectar secretion pattern: S331, Identification of peak nectar secretion period: This indicates the point in time when nectar secretion reaches its peak.

[0054] S332, High Nectar Secretion Duration: Sets the high nectar secretion threshold. Calculation satisfies Total length of continuous time period: ;in, The time interval between adjacent sampling points. Based on the statistical distribution of predicted nectar secretion values ​​in historical or experimental data, the 70%–80% percentile value is taken as the threshold.

[0055] S333, Nectar secretion rate analysis: rate of ascent: ; Decline rate: ; in, This indicates the start and end times of the nectar secretion curve. These respectively reflect the rate of nectar secretion initiation and the rate of decline at the end of secretion.

[0056] S33 quantitatively assesses the nectar secretion patterns of nectar-producing plants based on the morphological characteristics of their dynamic nectar secretion curves. Specifically, by identifying the peak position of the curve, the timing of the peak nectar secretion period is determined, thus identifying the key period of strongest nectar secretion capacity. A high nectar secretion threshold is set, and the total length of time the curve remains above this threshold is calculated to measure the plant's ability to maintain efficient nectar secretion throughout its flowering period. Finally, by analyzing the slopes of the rising and falling phases of the curve, the growth rate of the nectar secretion initiation phase and the decline rate of the nectar secretion decline phase are evaluated, reflecting the intensity of the initiation and the smoothness of the decline in the nectar secretion process. This transforms the dynamic nectar secretion process into quantifiable index parameters, thereby describing the rhythmicity and efficiency characteristics of nectar secretion behavior and providing a basis for nectar source selection, beekeeping timing, and ecological adaptability research. By deeply integrating spectral prediction values ​​with time series data and introducing curve analysis methods, a complete chain modeling from data points to physiological processes and management decisions is achieved.

[0057] S34, Total Honey Secretion Potential Assessment: The fitted curve is obtained by numerical integration. Throughout the flowering period Integrate the curve on the time axis and calculate the area under the curve, which is used as an indicator of total nectar secretion potential. ; Or, expressed in discrete form: ; in, The value represents the total nectar secretion potential, expressed in mg / flower. The area of ​​the integral region reflects an approximate estimate of the total nectar secretion during the entire flowering period, serving as a quantitative basis for assessing the quality of nectar source plants and subsequent beekeeping deployment.

[0058] In the assessment of total nectar secretion potential, a quantitative index reflecting the total nectar secretion capacity of nectar-producing plants, namely total nectar secretion potential, is calculated by integrating the dynamic nectar secretion curve over the entire flowering period. After obtaining the dynamic nectar secretion curve during the flowering period, the area enclosed by the curve and the time axis is used as the assessment object, with time as the horizontal axis and nectar secretion amount as the vertical axis. The size of this area represents the cumulative nectar secretion capacity of the nectar-producing plants throughout the entire flowering period.

[0059] In practice, numerical integration can be used for calculation. The trapezoidal method is employed to weight the predicted nectar secretion values ​​between adjacent time points and multiply by the time interval, thus gradually accumulating the total value for the entire flowering period. This integral result not only provides a quantitative reference for apiaries in selecting nectar sources and planning honey harvesting windows, but can also be used for comparative analysis of nectar source performance under different plants, regions, or climatic conditions. By combining spectral prediction results with time-series evaluation logic, the transformation of nectar secretion capacity from instantaneous characterization to periodic cumulative evaluation is realized, significantly improving the accuracy and practicality of nectar source evaluation.

[0060] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0061] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A rapid detection and dynamic evaluation method for nectar secretion from nectar-producing plants using near-infrared spectroscopy, characterized in that, Includes the following steps: S1. Determine the flowering stage of the target nectar source plant; the flowering stage includes at least the nectar secretion initiation stage, the nectar secretion peak stage, and the nectar secretion decline stage; use a near-infrared spectrometer to collect spectral data of the target nectar source plant flowers at the current flowering stage; S2. Input the spectral data into the stage-specific nectar secretion analysis model corresponding to the current flowering stage, and output the predicted nectar secretion value for the current flowering stage; the stage-specific nectar secretion analysis model is a prediction model independently established for different flowering stages; S3. During the target flowering period, repeat S1 to S2 at a predetermined frequency to obtain a time series consisting of multiple predicted nectar secretion values; generate a dynamic curve of nectar secretion during the flowering period of the target nectar source plant based on the time series; evaluate its nectar secretion pattern and total nectar secretion potential based on the morphological characteristics of the dynamic curve of nectar secretion during the flowering period.

2. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 1, characterized in that, The determination of the flowering stage includes acquiring a visible light image of the target flower, analyzing its corolla opening, petal color and morphological characteristics through an image recognition unit, and matching it with a database of flowering stage morphological characteristics of the target nectar source plant flower varieties to determine its flowering stage.

3. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 1, characterized in that, The acquisition of spectral data of the target nectar source plant flowers during their current flowering stage using a near-infrared spectrometer includes: Align the probe of the near-infrared spectrometer vertically with and lightly touch the nectary area or the base of the stamen to ensure that the position is consistent for each measurement. During the current flowering period, at fixed times each day, under conditions of avoiding direct sunlight, multiple spectral scans are performed on the same flower or the same batch of sample flowers, and the results of the multiple scans are averaged to obtain a spectral data representing the current flowering period.

4. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 1, characterized in that, The stage-specific nectar secretion analysis model is constructed using a partial least squares regression algorithm. It utilizes the correlation between nectar secretion and spectral data in samples at the current flowering stage to extract the principal spectral components related to the response variable, and establishes a linear regression model based on the principal spectral components.

5. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 4, characterized in that, S2 further includes preprocessing the spectral data by performing a standard normal transformation. The standard normal transformation is performed by centering the mean and normalizing the standard deviation of the spectral vector of each sample, and then inputting the processed data into the model.

6. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 5, characterized in that, S2 specifically includes: The preprocessed spectral data, based on the flowering stage identified in S1, is input into a stage-specific nectar secretion analysis model corresponding to the current flowering stage. The model outputs a quantitative nectar secretion prediction value, which characterizes the instantaneous nectar secretion capacity of the flower unit at the current flowering stage. The stage-specific nectar secretion analysis model is established independently for each different flowering stage as follows: Flower samples known to be in the nectar secretion initiation, peak nectar secretion, and nectar secretion decline stages were collected as the training set; The spectral data of each training set sample was collected using a near-infrared spectrometer, and its actual nectar secretion was measured simultaneously using the standard weight method to form a spectral-nectar secretion calibration dataset. For the calibration datasets of the initiation period, peak period, and decline period of nectar secretion, the partial least squares regression algorithm was used to train three distinct and optimized quantitative calibration models, namely the stage-specific nectar secretion analysis models.

7. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 6, characterized in that, The stage-specific nectar secretion analysis model also includes the introduction of a sparse regularization term to compress redundant band features during the regression coefficient solution process, retaining only the key spectral variables that contribute to the prediction of nectar secretion.

8. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 1, characterized in that, In S3, with a fixed 24-hour cycle, S1 to S2 are repeated on the same batch of pre-marked target nectar plant flowers at the same fixed time each day until the flowering period ends. The predicted nectar secretion value output after each execution is associated with the corresponding collection date and timestamp, and stored in chronological order to form a time-series sequence of predicted nectar secretion values ​​ordered by time.

9. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 8, characterized in that, The process of generating the dynamic curve of nectar secretion during the flowering period of the nectar-producing plant based on the time sequence specifically includes: In a coordinate system with time as the horizontal axis and the predicted nectar secretion value as the vertical axis, each data point in the time series is labeled; a curve fitting algorithm is used to connect all data points to generate a smooth and continuous dynamic change curve of nectar secretion during the flowering period, so as to intuitively show the evolution trend of nectar secretion with the flowering period.

10. The method for rapid detection and dynamic evaluation of nectar secretion from nectar-producing plants using near-infrared spectroscopy analysis according to claim 9, characterized in that, The assessment of nectar secretion patterns and total nectar secretion potential based on the morphological characteristics of the dynamic curve of nectar secretion during the flowering period specifically includes: Nectar secretion pattern assessment: Analyze the morphological characteristics of the curve, including identifying the peak point of the curve to determine the occurrence time of the peak secretion period, calculating the duration for which the peak of the curve exceeds the set high nectar secretion threshold to assess the duration of efficient nectar secretion, and analyzing the slope of the rising and falling branches of the curve to determine the rate of nectar secretion initiation and the speed of decline. Total nectar secretion potential assessment: The area of ​​the closed region enclosed by the dynamic curve of nectar secretion during flowering and the horizontal axis is calculated by numerical integration method, and the integrated area is used as a quantitative indicator to evaluate the total nectar secretion potential of the nectar source plant.

Images