Facility crop photosynthetic potential prediction method and system based on spectral characteristics

Abstract

The invention relates to a facility crop photosynthetic potential prediction method and system based on spectral characteristics, and the method comprises the steps: obtaining sample data through setting different illumination gradient culture environments, carrying out the conversion preprocessing of data through a standard normal variable, extracting a characteristic wavelength through employing a competitive adaptive reweighting algorithm and a continuous projection method, and carrying out the prediction of the photosynthetic potential of a facility crop. And establishing a prediction model by using a regression type support vector machine algorithm to predict the photosynthetic potential of the crops. The prediction system comprises a main control processing unit, a driving light source unit, a reflected light detection unit, a power supply unit, a communication unit, a control software unit, a display unit and the like. The main control processing unit controls operation of all parts and processes data, the light source unit is driven to provide detection light, the reflected light detection unit detects light intensity and converts and processes signals, the communication unit uploads data, and the control software and the display unit calculate and display results. According to the method, the photosynthetic potential of facility crops can be rapidly and accurately evaluated at low cost, a scientific basis is provided for agricultural production, and crop growth state monitoring is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of intelligent agricultural technology, and in particular to a method and system for predicting the photosynthetic potential of facility crops based on spectral characteristics. Background Art

[0002] Chlorophyll is the essential pigment for photosynthesis. When chlorophyll molecules absorb light energy, they transition from the ground state to an excited state and then back to the ground state. During this process, a portion of the energy is re-emitted as fluorescence, forming chlorophyll fluorescence. Measuring this fluorescence provides indirect insights into the processes involved in photosynthesis. Among the commonly used parameters for chlorophyll fluorescence, the maximum photochemical quantum yield (Fv / Fm) of PS II represents the plant's photosynthetic potential. This reflects the maximum light energy conversion efficiency of the PS II reaction center and is a key indicator in plant photosynthesis research. Real-time and non-destructive measurement of a plant's photosynthetic potential is crucial for characterizing its physiological state and its response to the external environment. However, the chlorophyll fluorescence instrumentation required for traditional Fv / Fm measurements is expensive and requires prolonged dark adaptation of the plant, limiting its application in practical agricultural production.

[0003] Spectroscopic analysis is an analytical technique based on the interaction between matter and light. It is sensitive, non-destructive, rapid, and efficient. To address the limitations of traditional Fv / Fm measurement methods for real-time measurement, numerous researchers have investigated the use of spectroscopy to predict Fv / Fm. The results demonstrate a linear relationship between leaf reflectance spectra and chlorophyll fluorescence parameters, demonstrating the feasibility of spectroscopy for fluorescence measurement.

[0004] However, most current studies use a modeling method based on vegetation index. This method contains fewer dimensions of spectral data and mostly uses simple fitting methods, which limits the accuracy of its Fv / Fm evaluation. On the other hand, the traditional spectral detection system consists of multiple modules, has high requirements for the stability of the measurement environment, and needs to be used with a PC to collect data. This configuration not only makes the equipment inconvenient to carry, but also the entire measurement process is relatively complicated, making it difficult to meet the needs of rapid detection in field or facility environments. Therefore, how to develop an efficient, accurate and convenient method for predicting the photosynthetic potential of facility crops based on spectral characteristics to meet the needs of rapid and accurate detection of the photosynthetic potential of facility crops in actual agricultural production is an urgent problem that needs to be solved. Summary of the Invention

[0005] In response to the technical problems existing in the prior art, the first purpose of the present invention is to provide a method for predicting the photosynthetic potential of facility crops based on spectral characteristics, which realizes non-destructive, rapid and low-cost detection of the photosynthetic potential of facility crops by integrating CARS-SPA characteristic wavelength screening, SNV data preprocessing and GA-SVR modeling.

[0006] The second purpose of the present invention is to provide a facility crop photosynthetic potential prediction system based on spectral characteristics, to achieve rapid, accurate and low-cost evaluation of the facility crop photosynthetic potential, and to provide a scientific basis for agricultural production.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for predicting the photosynthetic potential of greenhouse crops based on spectral characteristics comprises the following steps:

[0009] S1. Obtaining sample data: Setting up multiple cultivation environments with the same temperature, humidity, and carbon dioxide concentration under different light gradients. Waiting for differences in the crops under different light treatments, randomly select crop leaves as experimental samples and measure the photosynthetic potential Fv / Fm and visible-near infrared reflectance spectra of the crop leaves as sample data.

[0010] S2. Preprocessing the sample data: Using standard normal variable transformation to preprocess the data to eliminate the effects of sample surface particle size, surface scattering, and optical path length changes on the reflectance spectrum;

[0011] S3. Extract characteristic wavelengths: A competitive adaptive reweighting algorithm is first used to extract characteristic wavelength combinations that are beneficial to the accuracy of the prediction model. A continuous projection method is then used to screen out redundant variables with high repetitiveness. The training set and test set are then randomly divided into a 4:1 ratio to reduce model complexity and improve prediction accuracy.

[0012] S4. Using the reflected light intensity of the training set sample spectral characteristic wavelength as input and the crop Fv / Fm as output, a crop photosynthetic potential prediction model is established using a regression support vector machine algorithm;

[0013] S5. Use the photosynthetic potential prediction model to predict the photosynthetic potential of crops.

[0014] Furthermore, the steps of transforming the standard normal variable are:

[0015] Calculate the mean of each variable using the formula:

[0016]

[0017] in, represents the mean value of the jth wavelength point, n represents the number of samples, x ij Represents the spectrum value of the i-th sample at the j-th wavelength point;

[0018] Calculate the standard deviation of each variable using the formula:

[0019]

[0020] Among them, s j represents the standard deviation of the j-th wavelength point;

[0021] Perform SNV transformation, the calculation formula is:

[0022]

[0023] Among them, z ij It represents the spectrum value of the i-th sample at the j-th wavelength after SNV transformation.

[0024] Furthermore, the steps of the competitive adaptive reweighting algorithm are:

[0025] The Monte Carlo sampling method is used. Each time, 80% of the samples are randomly selected from the data set to enter the modeling set, and the remaining 20% are used as the prediction set to establish the PLS model. The number of sampling times N of the Monte Carlo method is set in advance, and the absolute value weight of the regression coefficient in the PLS model of each sampling process is recorded. The calculation formula is:

[0026]

[0027] Among them, b i is the regression coefficient of the i-th variable;

[0028] The exponential decay function (EDF) is used to forcibly remove wavelengths with relatively small absolute value weights of regression coefficients. When the PLS model is established based on Monte Carlo sampling for the i-th time, the proportion of retained wavelength points R is obtained according to the EDF. i , and its calculation formula is:

[0029]

[0030] Among them, the calculation formulas of μ and k are:

[0031]

[0032]

[0033] Where N is the number of sampling times, n is the number of original wavelength points;

[0034] The number of samples selected each time is R i *n wavelength variables, perform PLS modeling, and calculate the cross-validation root mean square error RMSECV value;

[0035] After N samplings are completed, the CARS algorithm obtains N groups of candidate characteristic wavelength subsets and the corresponding RMSECV values, and selects the wavelength variable subset corresponding to the minimum RMSECV value as the characteristic wavelength.

[0036] Furthermore, the steps of the continuous projection method are:

[0037] Select the initial wavelength and randomly select an initial wavelength from the spectrum matrix as the feature variable;

[0038] Perform projection calculation. For each remaining variable, calculate its projection on the selected feature variable set. The calculation formula is:

[0039]

[0040] Among them, x j is the variable currently under consideration, x k(n-1) is the feature variable selected in the previous iteration, P xj is x j In x k(n-1) projection on;

[0041] According to the projection result, the variable that maximizes the projected vector modulus is selected as the next feature variable:

[0042]

[0043] Among them, s is the index set of the feature variables that are not selected into the feature variable set, and k(n) is the index of the feature variable selected in this iteration;

[0044] Remove the selected feature variables from the original data set and update the set of unselected feature variables;

[0045] Repeat the above steps until the preset number of features is reached, and output the final selected feature variable set.

[0046] Furthermore, the photosynthetic potential prediction model comprises the following steps:

[0047] Read in modeling data and perform data normalization;

[0048] Select modeling parameters, including regularization parameter c, kernel function, and kernel function parameter g;

[0049] A regression support vector machine model was established based on the training set samples and optimal parameters to obtain a photosynthetic potential prediction model.

[0050] Furthermore, the data normalization process adopts a linear normalization method with an interval of [0,1], and the formula is:

[0051]

[0052] Among them, x * is the data after normalization, x is the data to be normalized, x max and x minare the maximum and minimum values in the data sequence respectively;

[0053] The kernel function uses the radial basis kernel function, and the genetic algorithm is used to find the optimal values of parameters c and g. The index range of c is set to [0.01, 10], and the number of genes is 6.

[0054] Set the index range of g to [0.01, 5], the number of genes to 5, and the encoding method to binary encoding. The encoding formula is:

[0055]

[0056] The decoding formula is:

[0057]

[0058] Among them, b is the encoded binary string, m is the number of binary string characters taken by the chromosome, a is the required encoded decimal number, a max is the maximum decimal number in the encoding space, a min The minimum decimal number in the encoding space.

[0059] A photosynthetic potential prediction system for greenhouse crops based on spectral characteristics, comprising:

[0060] A main control processing unit, which has a single-chip microcomputer and is used to control the switch of the driving light source unit, collect the reflected light signal of the reflected light detection unit, and transmit the processed data to the communication unit;

[0061] A light source driving unit comprising a constant current driving circuit and a light source control circuit, wherein the constant current driving circuit is used to provide a stable constant current for the light source, and the light source control circuit is used to control the light source to emit detection light;

[0062] A reflected light detection unit is used to detect the intensity of diffusely reflected light from the blades and perform signal conversion and processing. The unit includes a photoelectric conversion circuit, a signal amplification circuit, and an analog-to-digital conversion circuit. The photoelectric conversion circuit is used to convert the optical signal into an electrical signal, the signal amplification circuit is used to amplify the electrical signal, and the analog-to-digital conversion circuit is used to convert the analog signal into a digital signal that can be processed by the single-chip microcomputer.

[0063] A power supply unit, which is used to provide operating voltage for each unit;

[0064] A communication unit, which includes a wireless communication module and peripheral circuits. The wireless communication module supports Wi-Fi or Bluetooth protocols and is used to upload the collected data to the server database;

[0065] The control software and display unit are used to read data from the server database, call the crop photosynthetic potential prediction model to calculate the Fv / Fm value, store the test results and send them to the mobile terminal for display.

[0066] Furthermore, the single-chip microcomputer is an STM32F103 single-chip microcomputer, which is respectively communicated with the constant current drive circuit and the analog-to-digital conversion circuit through a standard I / O interface, and is connected to the wireless communication module through a USART module.

[0067] Furthermore, the driving light source unit and the reflected light detection unit are arranged in a reflected light detection probe, and the reflected light detection probe includes a shell with an open top, and a plurality of LED lights are circumferentially arranged on the inner wall of the lower part of the shell. The light directions of the plurality of LED lights converge at a certain angle toward the center position of the top of the shell, and then illuminate the blade to be detected. A photoelectric sensor is provided at the bottom of the shell, which is used to receive the reflected light formed after the LED light illuminates the blade to be detected.

[0068] Furthermore, a light shielding platform is provided at the bottom of the shell for blocking stray light emitted by the LED lamp from interfering with the photoelectric sensor.

[0069] The present invention has the following advantages:

[0070] 1. The prediction method of the present invention abandons the traditional method of measuring Fv / Fm by relying on a chlorophyll fluorescence meter. The traditional method not only requires the plants to undergo long periods of dark adaptation, but also the equipment is expensive, which limits its application in actual agricultural production. The present invention constructs a prediction model based on the spectral characteristics of crops, and directly predicts the photosynthetic potential by analyzing the reflected light intensity of specific wavelengths in the plant reflectance spectrum. It avoids the cumbersome dark adaptation process, shortens the detection time, and reduces the cost of detection equipment, improves detection efficiency, and makes the detection work more economical. In data processing and model construction, on the one hand, the sample data is pre-processed using a standard normal variable transformation, which effectively eliminates the interference of factors such as sample surface particle size, surface scattering, and optical path changes on the reflectance spectrum, greatly improving the data quality and providing a strong guarantee for the accuracy of subsequent modeling analysis. On the other hand, by combining the competitive adaptive reweighting algorithm (CARS) and the successive projection method (SPA) to extract characteristic wavelengths, while ensuring the accuracy of the model, the number of characteristic wavelengths is reduced as much as possible, the model structure is simplified, and the prediction speed of the model is further improved. In addition, when using the regression support vector machine algorithm (SVR) to establish a crop photosynthetic potential prediction model, the data is first normalized to unify the data order of magnitude to avoid sample imbalance problems; then the genetic algorithm (GA) is used to optimize the SVR model's regularization parameter c, kernel function, and kernel function parameter g to achieve optimal model performance, significantly improving the accuracy and reliability of Fv / Fm photosynthetic potential prediction.

[0071] 2. The prediction system of the present invention integrates multiple functional modules. The main control processing unit, based on an STM32F103 single-chip microcontroller, controls each unit and processes transmitted data. The communication unit uploads collected data to a cloud server database. The control software and display unit reads data from the database, invokes a prediction model to calculate the Fv / Fm value, then stores the result and sends it to a mobile device for display. This enables rapid, non-destructive, and low-cost intelligent detection of crop Fv / Fm. The light source control circuit driving the light source unit utilizes a six-way ring-shaped narrow-bandwidth LED array. These LEDs are arranged at specific angles within the reflected light detection probe. When operating, they uniformly illuminate the crop leaves, providing a stable and uniform light source for detection. The reflected light detection unit, also located within the reflected light detection probe, utilizes photoelectric conversion, signal amplification, and analog-to-digital conversion circuits to accurately and effectively detect the intensity of diffusely reflected light from the leaves. To ensure the accuracy and high quality of the detection results, a narrow-bandwidth LED light source and photodiode are used, effectively improving detection precision. Furthermore, a light shield is installed within the reflected light detection probe to block stray light generated by the LED itself and the surrounding environment, preventing it from interfering with the detection signal. Furthermore, a first-order low-pass filter circuit eliminates high-frequency noise, further purifying the signal. This simplifies the detection structure and significantly reduces costs, facilitating the device's portability and widespread adoption. It effectively meets the need for rapid detection of crop photosynthetic potential in field or facility environments, providing strong technical support for agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 This is a flow chart of the method for predicting photosynthetic potential of facility crops based on spectral characteristics of the present invention.

[0073] Figure 2 The present invention is a flow chart of establishing a photosynthetic potential prediction model by using a GA-SVR algorithm in the method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics.

[0074] Figure 3 It is a comparison chart of the predicted value and the true value of the Fv / Fm prediction model of the present invention.

[0075] Figure 4 This is a control flow chart of the system for predicting photosynthetic potential of facility crops based on spectral characteristics of the present invention.

[0076] Figure 5 It is a structural schematic diagram of the greenhouse crop photosynthetic potential prediction system based on spectral characteristics of the present invention.

[0077] Figure 6 It is a structural schematic diagram of the greenhouse crop photosynthetic potential prediction system based on spectral characteristics of the present invention.

[0078] Figure 7This is a workflow diagram of the system for predicting photosynthetic potential of facility crops based on spectral characteristics of the present invention.

[0079] Among them, 1 is the reflected light detection probe, 101 is the shell, 101a is the light shielding platform, 2 is the driving light source unit, 201 is the LED lamp, 202 is the constant current driving circuit, 203 is the light source control circuit, 3 is the reflected light detection unit, 301 is the photoelectric sensor, 302 is the photoelectric conversion circuit, 303 is the signal amplification circuit, 304 is the analog-to-digital conversion circuit, 4 is the main control processing unit, 5 is the power supply unit, 6 is the communication unit, 601 is the wireless communication module, 7 is the control software and display unit, 701 is the photosynthetic potential prediction model, 702 is the server, and 703 is the mobile terminal. DETAILED DESCRIPTION

[0080] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be embodied in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art. Like reference numerals in the drawings represent like or similar parts, and thus repetitive description thereof will be omitted.

[0081] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner.In the following description, many specific details are provided so as to provide a full understanding of the embodiments of the present application. However, it will be appreciated by those skilled in the art that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps etc. can be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0082] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0083] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all promotional information and operations / steps, nor do they necessarily require execution in the order described. For example, some operations / steps may be broken down, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0084] It should be understood that although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Thus, the first component discussed below can be referred to as the second component without departing from the teachings of the concepts of the present application. As used herein, the term "and / or" and similar terms include all combinations of any, multiple, and all of the associated listed items.

[0085] Those skilled in the art will understand that the drawings are merely schematic diagrams of example embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing the present application, and therefore cannot be used to limit the scope of protection of the present application.

[0086] Before introducing the embodiments of the present application in detail, some of the terms involved in the embodiments of the present application are first explained to facilitate understanding by those skilled in the art.

[0087] The photosynthetic potential (Fv / Fm), or the maximum photochemical quantum yield of PSII, is a key metric in plant photosynthesis research and is used to measure a plant's photosynthetic potential. In photosynthesis, chlorophyll, a fundamental pigment, absorbs light energy and transitions from its ground state to an excited state, then returns to the ground state, emitting some of the energy as chlorophyll fluorescence. Fv / Fm reflects the maximum light-energy conversion efficiency of the PSII reaction center. Its calculation formula is Fv / Fm = (Fm - Fo) / Fm, where Fo is the initial fluorescence, representing the fluorescence yield when the PSII reaction center is fully open; Fm is the maximum fluorescence, representing the fluorescence yield when the PSII reaction center is fully closed. A larger Fv / Fm value indicates a higher light-energy conversion efficiency of the PSII reaction center, a greater ability of the plant to convert light energy into chemical energy, and, consequently, a higher photosynthetic potential. Real-time, non-destructive testing of plant Fv / Fm helps accurately assess a plant's physiological state and understand its response to the external environment, and is of great significance in agricultural production and plant research.

[0088] The Standard Normal Variate transformation (SNV) algorithm is a method commonly used in spectral data preprocessing. Its purpose is to eliminate the influence of physical property differences of samples (such as surface particle size, surface scattering and optical path changes) on spectral reflectance, improve data quality, and enhance the accuracy of subsequent modeling analysis.

[0089] The competitive adaptive reweighted sampling (CARS) algorithm is a feature variable selection method that combines Monte Carlo sampling with the regression coefficients of the PLS model. It aims to optimize the prediction accuracy of the calibration model by iteratively building the prediction model to obtain the feature variables that are most suitable for improving model accuracy. The competitive adaptive reweighted sampling algorithm is a feature variable selection method that combines Monte Carlo sampling with the regression coefficients of the PLS model. It aims to optimize the prediction accuracy of the calibration model by progressively evaluating, analyzing, screening, and eliminating each wavenumber point in the spectrum, making it suitable for high-dimensional spectral screening.

[0090] The Successive Projections algorithm (SPA) is a forward-loop selection method that uses vector projection analysis to extract wavelength combinations with minimal redundancy and collinearity. This forward feature variable selection method starts with the first wavenumber and incorporates a new wavenumber at each iteration until the set wavenumber is reached. Its characteristic is that it selects wavenumber combinations with minimal redundancy and information content, thus addressing collinearity.

[0091] The Support Vector Regression (SVR) algorithm is an extended application of the Support Vector Machine (SVM) to regression problems and is a supervised learning algorithm. It aims to find an optimal regression function to predict continuous-valued outputs as accurately as possible. The core idea of SVR is to map the data in the input space to a high-dimensional feature space through a nonlinear mapping, and then find an optimal hyperplane in this high-dimensional space to fit the data. To deal with data noise and errors, SVR introduces an insensitive loss function (usually an ε-insensitive loss function), which allows a certain error range (ε) between the predicted value and the true value. As long as the error is within this range, the prediction is considered accurate and no loss is incurred.

[0092] First refer to Figure 1 Taking grapes as an example, the following is a specific embodiment of the method for predicting the photosynthetic potential of facility crops based on spectral characteristics.

[0093] S1. Obtain sample data

[0094] Under the same culture environment of temperature, humidity and carbon dioxide concentration, six light intensity gradients were set to cultivate grapes, with the light quantum flux density being 90, 140, 220, 280 and 340 μmol·m -2 ·s -1When the grape leaves under different light treatments showed differences, leaves were randomly selected as experimental samples, and the photosynthetic potential Fv / Fm and visible-near infrared reflectance spectra of the crop leaves were measured as sample data.

[0095] S2. Preprocess the sample data

[0096] The standard normal variable transformation algorithm is an effective spectral data preprocessing method that can eliminate interference factors such as scattering effects and baseline drift in the data and improve the data modeling performance. The standard normal variable transformation is used to preprocess the sample data. The steps are as follows:

[0097] Calculate the mean of each variable: For the collected spectral data matrix, calculate the average value of each wavelength point on all samples. The calculation formula is:

[0098]

[0099] in, represents the mean value of the jth wavelength point, n represents the number of samples, x ij Represents the spectral value of the i-th sample at the j-th wavelength point.

[0100] Calculate the standard deviation of each variable: Calculate the standard deviation of each wavelength point over all samples using the following formula:

[0101]

[0102] Among them, s j Represents the standard deviation of the j-th wavelength point.

[0103] Perform SNV transformation: Perform standard normal variable transformation on each wavelength point of each sample. The calculation formula is:

[0104]

[0105] Among them, z ij It represents the spectrum value of the i-th sample at the j-th wavelength after SNV transformation.

[0106] S3. Extract characteristic wavelength

[0107] Leaf reflectance spectra exhibit high linear repeatability and high levels of redundant information, leading to complex models and reduced accuracy. Characteristic wavelength screening can reduce the number of wavelength variables and improve model prediction speed. Therefore, a competitive adaptive reweighting algorithm is first used to extract characteristic wavelength combinations that contribute to model accuracy. Successive projections are then used to filter out highly reproducible, redundant variables.

[0108] The competitive adaptive reweighting algorithm first uses the Monte Carlo sampling method, randomly selecting 80% of the samples from the data set to enter the modeling set each time, and the remaining 20% is used as the prediction set to establish the PLS model. The number of Monte Carlo sampling times N is set in advance. The absolute value weight of the regression coefficient in the PLS model during each sampling process is recorded, and its calculation formula is:

[0109]

[0110] Among them, b i is the regression coefficient of the ith variable.

[0111] The exponential decay function (EDF) is used to forcibly remove wavelengths with relatively small absolute value weights of regression coefficients. When the PLS model is established based on Monte Carlo sampling for the i-th time, the proportion of retained wavelength points R is obtained according to the EDF. i , and its calculation formula is:

[0112]

[0113] Among them, the calculation formulas of μ and k are:

[0114]

[0115] Where N is the number of sampling times and n is the number of original wavelength points.

[0116] The number of samples selected each time is R i *n wavelength variables, perform PLS modeling, and calculate the cross-validation root mean square error RMSECV value.

[0117] After N samplings are completed, the CARS algorithm obtains N groups of candidate characteristic wavelength subsets and the corresponding RMSECV values, and selects the wavelength variable subset corresponding to the minimum RMSECV value as the characteristic wavelength.

[0118] The continuous projection method first selects the initial wavelength and randomly selects an initial wavelength from the spectrum matrix as the feature variable;

[0119] Perform projection calculation. For each remaining variable, calculate its projection on the selected feature variable set. The calculation formula is:

[0120]

[0121] Among them, x j is the variable currently under consideration, x k(n-1) is the feature variable selected in the previous iteration, P xj is x j In x k(n-1) projection on;

[0122] According to the projection result, the variable that maximizes the projected vector modulus is selected as the next feature variable:

[0123]

[0124] Among them, s is the index set of the feature variables that are not selected into the feature variable set, and k(n) is the index of the feature variable selected in this iteration;

[0125] Remove the selected feature variables from the original data set and update the set of unselected feature variables;

[0126] Repeat the above steps until the preset number of features is reached, and output the final selected feature variable set.

[0127] By combining CARS and SPA, it is possible to minimize the number of characteristic wavelengths while taking model accuracy into consideration.

[0128] S4. Establishing a photosynthetic potential prediction model using GA-SVR

[0129] Reference Figure 2 , which shows the overall process of establishing a photosynthetic potential prediction model using the GA-SVR algorithm. The establishment of the crop photosynthetic potential prediction model includes the following steps:

[0130] Normalize the data so that the data of each dimension are at the same order of magnitude to avoid sample imbalance caused by large data differences, which may cause the final model to deviate from the accurate hyperplane. The normalization operation uses the linear normalization method with an interval of [0,1] and the formula is:

[0131]

[0132] x * is the data after normalization, x is the data to be normalized, x max and x min are the maximum and minimum values in the data series, respectively.

[0133] The radial basis kernel function is used as the kernel function, and the genetic algorithm is used to find the optimal values of parameters c and g. The index range of c is set to [0.01, 10] and the number of genes is 6.

[0134] Set the index range of g to [0.01, 5], the number of genes to 5, and the encoding method to binary encoding. The encoding formula is:

[0135]

[0136] The decoding formula is:

[0137]

[0138] Among them, b is the encoded binary string, m is the number of binary string characters taken by the chromosome, a is the required encoded decimal number, a max is the maximum decimal number in the encoding space, a min The minimum decimal number in the encoding space.

[0139] Using the decoded c and g as key parameters, a regression support vector machine was trained using the training set. The chromosome fitness was assessed by calculating the coefficient of determination of the test set data, and a roulette wheel mechanism was used to eliminate and select individuals from the population. In the resulting offspring, the individuals with the highest fitness from the parent generation were retained, while those with the lowest fitness were replaced. The offspring then evolved further through crossover, mutation, and selection, ultimately iterating to the optimal solution for parameters c and g. A coefficient of variation of 0.2 and a crossover coefficient of 0.8 were selected, and 50 generations of iterative evolution were performed to obtain the optimal solution for parameters c and g.

[0140] Based on the training set samples and the optimal parameters, the SVR model is established to obtain the photosynthetic potential model f(x)=w T x+b, where w is the regression coefficient matrix and b is the bias vector.

[0141] Introducing a slack variable ζ into the model i To form a soft margin, the regression support vector machine can be transformed into:

[0142]

[0143] Among them, y i is the Fv / Fm of the i-th sample, ε is the insensitive loss function, ζ i and is the slack variable, and m is the total number of model training samples. By introducing Lagrange multipliers and taking the partial derivative of the Lagrange function in the above equation, we can transform it into the dual problem of a regression support vector machine and solve the hyperplane according to the KKT condition.

[0144] Validation of the Fv / Fm photosynthetic potential prediction model

[0145] The Fv / Fm photosynthetic potential prediction model was verified using the test set data. Figure 3 As shown, the coefficient of determination R between the predicted value and the actual value 2 =0.93, root mean square error RMSE =0.021, indicating that the model has high prediction accuracy.

[0146] To test the robustness of the prediction model, the experiment was repeated under fluctuating conditions of temperature (20-30°C) and humidity (50-70% RH) and the model performance was tested. The results are shown in Table 1. The coefficient of determination R of the model under different conditions is 2It is stable at 0.90-0.93, and the root mean square error RMSE ≤ 0.025, indicating that it has strong environmental adaptability.

[0147] Table 1 shows the model performance under different environmental conditions.

[0148]

[0149] Reference Figure 4 , which shows the control flow of the greenhouse crop photosynthetic potential prediction system based on spectral characteristics. It mainly includes a main control processing unit 4, a driving light source unit 2, a reflected light detection unit 3, a power supply unit 5, a communication unit 6, and a control software and display unit 7.

[0150] The main control processing unit 4 has a single-chip microcomputer, which is used to control the switch of the light source unit 2, collect the reflected light signal of the reflected light detection unit 3, and transmit the processed data to the communication unit 6. The single-chip microcomputer is an STM32F103 single-chip microcomputer, which is connected to the constant current drive circuit 202 and the analog-to-digital conversion circuit 304 through standard I / O interfaces, and is connected to the wireless communication module 601 through a USART module.

[0151] The light source driving unit 2 includes a constant current drive circuit 202 and a light source control circuit 203. The constant current drive circuit 202 provides a stable constant current for the light source, while the light source control circuit 203 controls the light source's emission of detection light. The constant current drive circuit 202 uses a linear LED driver chip to provide a stable constant current to meet the requirements of DC LED drive. The light source control circuit 203 uses a six-channel ring-shaped narrow-bandwidth LED array to emit detection light centered on the characteristic wavelength toward the crop leaves. MOS transistors are used to turn each LED on and off.

[0152] Among them, LED lamps are typical constant current devices that use DC drive mode, so the linear LED driver chip CN5501 is used to provide it with a stable constant current. The CN5501 output current calculation formula is:

[0153]

[0154] Among them, I LED is the output current, R CS is the resistance between the CS pin of CN5501 and ground.

[0155] Light source control circuit 203 can use an N-channel MOS transistor to implement LED switching. When the gate-source voltage of the MOS transistor is 0, there is no conductive channel between its drain and source, and it can be considered an open circuit. When the gate-source voltage exceeds the MOS transistor's conduction voltage, a conductive channel forms between its drain and source. Each MOS transistor gate is connected in series with a resistor and then connected to the core processor's I / O port. When the I / O port outputs a high level, its drain and source are conductive, and the current generated by the constant current drive circuit 202 flows into the light source.

[0156] Reflected light detection unit 3 detects the intensity of diffusely reflected light from the leaves and performs signal conversion and processing. It includes a photoelectric conversion circuit 302, a signal amplification circuit 303, and an analog-to-digital conversion circuit 304. Photoelectric conversion circuit 302 converts the optical signal into an electrical signal, while signal amplification circuit 303 amplifies the electrical signal. Analog-to-digital conversion circuit 304 converts the analog signal into a digital signal suitable for processing by the microcontroller. Photoelectric conversion circuit 302 utilizes a photodiode as a photoelectric sensor, characterized by high sensitivity, good high-frequency performance, compact size, and low cost. Combined with a narrow-wavelength, bandwidth-sensitive light source, this sensor meets the design requirements of this detection device.

[0157] Specifically, a first-order low-pass filter circuit can be used to eliminate high-frequency noise from the power supply signal. A photodiode is exposed to a 5V reverse voltage. When illuminated by light, it generates a photocurrent that flows through a 1kΩ sampling resistor. As the light intensity increases, the photocurrent increases, and the voltage across the sampling resistor increases, thus achieving optical-to-electrical signal conversion.

[0158] The signal amplification circuit 303 is a proportional operation circuit, and its signal gain coefficient is calculated as follows:

[0159]

[0160] Among them, R f The resistance is the feedback resistor, and R is the external resistor at the reverse end.

[0161] Specifically, signal amplification can be achieved by using an operational amplifier. In the signal amplification circuit 303, the voltage signal across the sampling resistor is input through the same-direction terminal of the operational amplifier, amplified and output through the output terminal, and then decoupled by a capacitor and transmitted to the analog-to-digital conversion circuit 304.

[0162] The analog-to-digital conversion circuit 304 uses an analog-to-digital conversion chip to convert the voltage signal, and the voltage signal is transmitted to the STM32 core processor after analog-to-digital conversion.

[0163] like Figure 5 and Figure 6As shown, it shows a preferred embodiment in which the driving light source unit 2 and the reflected light detection unit 3 are arranged in the reflected light detection probe 1, and the reflected light detection probe 1 includes a shell 101 with an open top, and the top opening corresponds to the target area of the blade to be detected. A photoelectric sensor 301 as a detection element is placed at the bottom center of the shell 101, which is used to receive the reflected light formed after the blade to be detected is irradiated by the LED lamp 201. In order to ensure that the received signal comes only from the target surface and avoid interference from other sources such as scattered light generated by the LED lamp 201 itself or the surrounding environment, a light shielding platform 101a is also provided at the bottom of the shell 101. The light shielding platform 101a is arranged around the outer side of the upper part of the photoelectric sensor 301, and can block the stray light emitted by the LED lamp 201 from entering the photoelectric sensor 301.

[0164] Multiple LED lights 201 are circumferentially arranged on the inner sidewall of the lower portion of the housing 101. Specifically, six LED lights 201, each emitting light of varying wavelengths, are arranged in a circular pattern, with a 60° interval between adjacent LED lights 201. This design ensures uniform illumination of the target area. The illumination from these LED lights 201 converges at an angle toward the center of the top of the housing 101, with the illumination angle of the LED lights 201 oriented at a 45° angle to the blade being inspected.

[0165] Power supply unit 5 provides operating voltage for each unit. Specifically, a lithium battery can be used as the power source. Lithium batteries have high rated voltage, high energy density, compact size, light weight, long life, high power handling capacity, and strong temperature adaptability. They are widely used in small electronic devices and precision instruments, making them an ideal power source for this device. Furthermore, to meet the operating requirements of the hardware circuit, a linear regulator can be used to generate different stable voltages. Capacitors are connected to the voltage input and output terminals of the linear regulator to enhance voltage signal stability.

[0166] The communication unit 6 includes a wireless communication module 601 and peripheral circuits, which are used to upload the collected reflected light data to the database of the server 702. Specifically, the wireless communication module 601 (Wi-Fi module) can adopt the ESP8266 series module, and the STM32 core processor uses the USART transceiver module to send the collected light intensity data to the ESP8266 chip through the RX and TX ports.

[0167] The control software and display unit 7 is used to read data from the server 702 database, call the crop photosynthetic potential prediction model 701 to calculate the Fv / Fm value, store the test results, and send them to the mobile terminal 703 for display. Specifically, the control software and display unit 7 reads the reflected light data from the server 702 database, then calls the prediction model to calculate the Fv / Fm value corresponding to the reflected light data, stores the test results in the database, sends them to the mobile terminal, and finally displays the test results.

[0168] The workflow of the greenhouse crop photosynthetic potential prediction system based on spectral characteristics includes:

[0169] The power supply unit 5 provides a stable operating voltage to each unit of the system, and the system starts;

[0170] The STM32F103 microcontroller of the main control processing unit 4 initializes each I / O interface;

[0171] The main control processing unit 4 controls the light source control circuit 203 to sequentially turn on the LED lights. Each LED light is turned on for 300ms. During this time, the reflected light detection unit collects, converts, and processes the reflected light signal. Specifically, the photoelectric conversion circuit 302 converts the optical signal into an electrical signal, the signal amplification circuit 303 amplifies the electrical signal, and the analog-to-digital conversion circuit 304 converts the analog signal into a digital signal and transmits it to the STM32 core processor.

[0172] Repeat step 3 until data collection is completed for all 6 LED lights;

[0173] The STM32 core processor uploads the stored reflected light data to the server 702 database via the ESP8266 module of the communication unit 6;

[0174] The control software and display unit 7 reads data from the server 702 database, calls the crop photosynthetic potential prediction model 701 to calculate the Fv / Fm value, stores the test results and sends them to the mobile terminal 703 for display.

[0175] It should be noted that the hardware design described above serves only as a specific example to illustrate the specific structure and appearance of the greenhouse crop photosynthetic potential detection system provided by the present invention, the installation locations of each module, and the specific hardware selection for each module in actual application. The system's software configuration should be compatible with the hardware design. The specific hardware and software design settings can be adjusted according to actual needs and are not limited in this embodiment.

[0176] The proposed spectral-based photosynthetic potential prediction method and detection system for greenhouse crops can be used to accurately assess the physiological status of crops. By utilizing specific spectral characteristic wavelengths for detection, this method not only reduces costs but also increases detection speed, better meeting practical needs and demonstrating high practicality and convenience.

[0177] The embodiments and examples set forth herein are provided to best illustrate embodiments according to the present application and its specific applications, and thereby enable those skilled in the art to make and use the present application. However, those skilled in the art will appreciate that the above description and examples are provided for ease of illustration and example only. The descriptions set forth are not intended to be exhaustive of all aspects of the present application or to limit the present application to the precise forms disclosed.

Claims

1. A method for predicting the photosynthetic potential of greenhouse crops based on spectral characteristics, characterized in that: The following steps are involved: S1. Obtaining sample data: Setting up multiple cultivation environments with the same temperature, humidity, and carbon dioxide concentration under different light gradients. Waiting for differences in the crops under different light treatments, randomly select crop leaves as experimental samples and measure the photosynthetic potential Fv / Fm and visible-near infrared reflectance spectra of the crop leaves as sample data. S2. Preprocessing the sample data: Using standard normal variable transformation to preprocess the data to eliminate the effects of sample surface particle size, surface scattering, and optical path length changes on the reflectance spectrum; S3. Extract characteristic wavelengths: First, a competitive adaptive reweighting algorithm is used to extract characteristic wavelength combinations that are beneficial to the accuracy of the prediction model. Then, a continuous projection method is used to screen out redundant variables with high repeatability. The training set and test set are then randomly divided into 4:1 parts. S4. Using the reflected light intensity of the training set sample spectral characteristic wavelength as input and the crop Fv / Fm as output, a crop photosynthetic potential prediction model is established using a regression support vector machine algorithm; S5. Use the photosynthetic potential prediction model to predict the photosynthetic potential of crops.

2. The method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 1, characterized in that: The steps of standard normal variable transformation are: Calculate the mean of each variable using the formula: in, represents the mean value of the jth wavelength point, n represents the number of samples, x ij Represents the spectrum value of the i-th sample at the j-th wavelength point; Calculate the standard deviation of each variable using the formula: Among them, s j represents the standard deviation of the j-th wavelength point; Perform SNV transformation, the calculation formula is: Among them, z ij It represents the spectrum value of the i-th sample at the j-th wavelength after SNV transformation.

3. The method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 1, characterized in that: The steps of the competitive adaptive reweighting algorithm are: The Monte Carlo sampling method is used. Each time, 80% of the samples are randomly selected from the data set to enter the modeling set, and the remaining 20% are used as the prediction set to establish the PLS model. The number of sampling times N of the Monte Carlo method is set in advance, and the absolute value weight of the regression coefficient in the PLS model of each sampling process is recorded. The calculation formula is: Among them, b i is the regression coefficient of the i-th variable; The exponential decay function (EDF) is used to forcibly remove wavelengths with relatively small absolute value weights of regression coefficients. When the PLS model is established based on Monte Carlo sampling for the i-th time, the proportion of retained wavelength points R is obtained according to the EDF. i , and its calculation formula is: Among them, the calculation formulas of μ and k are: Where N is the number of sampling times, n is the number of original wavelength points; The number of samples selected each time is R i *n wavelength variables, perform PLS modeling, and calculate the cross-validation root mean square error RMSECV value; After N samplings are completed, the CARS algorithm obtains N groups of candidate characteristic wavelength subsets and the corresponding RMSECV values, and selects the wavelength variable subset corresponding to the minimum RMSECV value as the characteristic wavelength.

4. The method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 1, characterized in that: The steps of the continuous projection method are: Select the initial wavelength and randomly select an initial wavelength from the spectrum matrix as the feature variable; Perform projection calculation. For each remaining variable, calculate its projection on the selected feature variable set. The calculation formula is: Among them, x j is the variable currently under consideration, x k(n-1) is the feature variable selected in the previous iteration, P xj is x j In x k(n-1) projection on; According to the projection result, the variable that maximizes the projected vector modulus is selected as the next feature variable: Among them, s is the index set of the feature variables that are not selected into the feature variable set, and k(n) is the index of the feature variable selected in this iteration; Remove the selected feature variables from the original data set and update the set of unselected feature variables; Repeat the above steps until the preset number of features is reached, and output the final selected feature variable set.

5. The method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 1, characterized in that: The photosynthetic potential prediction model comprises the following steps: Read in modeling data and perform data normalization; Select modeling parameters, including regularization parameter c, kernel function, and kernel function parameter g; A regression support vector machine model was established based on the training set samples and optimal parameters to obtain a photosynthetic potential prediction model.

6. The method for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 5, characterized in that: The data normalization process adopts a linear normalization method with an interval of [0,1], and the formula is: Among them, x * is the data after normalization, x is the data to be normalized, x max and x min are the maximum and minimum values in the data sequence respectively; The kernel function uses the radial basis kernel function, and the genetic algorithm is used to find the optimal values of parameters c and g. The index range of c is set to [0.01, 10], and the number of genes is 6. Set the index range of g to [0.01, 5], the number of genes to 5, and the encoding method to binary encoding. The encoding formula is: The decoding formula is: Among them, b is the encoded binary string, m is the number of binary string characters taken by the chromosome, a is the required encoded decimal number, a max is the maximum decimal number in the encoding space, a min The minimum decimal number in the encoding space.

7. A system for predicting photosynthetic potential of greenhouse crops based on spectral characteristics, characterized in that: include: A main control processing unit, which has a single-chip microcomputer and is used to control the switch of the driving light source unit, collect the reflected light signal of the reflected light detection unit, and transmit the processed data to the communication unit; A light source driving unit comprising a constant current driving circuit and a light source control circuit, wherein the constant current driving circuit is used to provide a stable constant current for the light source, and the light source control circuit is used to control the light source to emit detection light; A reflected light detection unit is used to detect the intensity of diffusely reflected light from the blades and perform signal conversion and processing. The reflected light detection unit includes a photoelectric conversion circuit, a signal amplification circuit, and an analog-to-digital conversion circuit. The photoelectric conversion circuit is used to convert the light signal into an electrical signal, the signal amplification circuit is used to amplify the electrical signal, and the analog-to-digital conversion circuit is used to convert the analog signal into a digital signal that can be processed by the single-chip microcomputer. A power supply unit is used to provide operating voltage for each unit. A communication unit, which includes a wireless communication module and peripheral circuits, and is used to upload the collected data to a server database; The control software and display unit are used to read data from the server database, call the crop photosynthetic potential prediction model to calculate the Fv / Fm value, store the test results and send them to the mobile terminal for display.

8. The system for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 7, characterized in that: The single-chip microcomputer is an STM32F103 single-chip microcomputer, which is respectively communicated with the constant current drive circuit and the analog-to-digital conversion circuit through a standard I / O interface, and is connected to the wireless communication module through a USART module.

9. The system for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 7, characterized in that: The driving light source unit and the reflected light detection unit are arranged in the reflected light detection probe, and the reflected light detection probe includes a shell with an open top, and a plurality of LED lights are circumferentially arranged on the inner wall of the lower part of the shell. The light directions of the plurality of LED lights converge at a certain angle toward the center position of the top of the shell, and then illuminate the blade to be detected. A photoelectric sensor is provided at the bottom of the shell, which is used to receive the reflected light formed after the blade to be detected is illuminated by the LED light.

10. The system for predicting photosynthetic potential of greenhouse crops based on spectral characteristics according to claim 9, characterized in that: The bottom of the housing is also provided with a light shielding platform for preventing stray light emitted by the LED lamp from interfering with the photoelectric sensor.

Images