Method for detecting tomato root rot based on volatile group information

Abstract

The invention relates to the technical field of tomato disease control, in particular to a method for detecting tomato root rot based on volatilization group information. The method comprises the four steps of volatile gas collection, data analysis, intelligent diagnosis and result display, and cloud database and model updating is assisted. The method comprises the following steps: firstly, obtaining tomato root volatile compounds through high-sensitivity sampling equipment, and analyzing volatile components by utilizing a GC-IMS technology to obtain lt; component, concentration gt; the method comprises the following steps: extracting data, generating a feature vector, combining generation and verification of an auditing identifier to ensure the compliance and integrity of the extracted data, accurately judging a disease grade based on a classification model, combining a health score and a classification result, providing prevention and treatment suggestions, and presenting a diagnosis result in a visual mode; and the cloud database continuously accumulates historical data, optimizes the classification model, and improves the system adaptability and the detection effect. According to the invention, early-stage accurate diagnosis and intelligent management of tomato root rot are realized, and the agricultural disease detection efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of tomato disease prevention and control, and in particular to a method for detecting tomato root rot based on volatile group information. Background Art

[0002] Tomato is an important economic crop worldwide and is widely planted in various agricultural production areas. Its yield and quality are directly related to the economic benefits of agriculture. However, tomatoes are extremely susceptible to diseases during their growth process. Root rot is a common and serious soil-borne disease caused by fungi, bacteria or other pathogens. It can cause root rot and weak growth of the plant, ultimately affecting fruit yield and even plant death. The disease is characterized by its hidden spread and rapid outbreak. Once it occurs, it is difficult to control in time and may cause serious economic losses.

[0003] A Chinese invention patent with announcement number CN117969771B discloses a method for monitoring tomato plant diseases, which specifically includes: initializing a tomato planting greenhouse scene, arranging a gas detector in the scene, using the gas detector to measure the concentration, and obtaining concentration characteristic values ​​in combination with various measurement results, and then performing nitrogen abnormality analysis through the concentration characteristic values ​​obtained through continuous measurement, and obtaining nitrogen abnormality characteristics, and finally performing disease monitoring and early warning based on the obtained nitrogen abnormality characteristics; effectively identifying the regularity of concentration changes in the greenhouse, and quantifying the degree of abnormal behavior of nitrogen dioxide emissions in the tomato planting greenhouse scene, thereby providing a reliable mathematical basis for analyzing and distinguishing non-infectious diseases caused by harmful gases from infectious diseases caused by pathogens, so as to more scientifically and efficiently distinguish tomato plants showing white spots, water-soaked spots, wilting, falling flowers and fruits, and thus enhancing the prevention and control efficiency and quality.

[0004] However, the current research on disease detection based on volatile group information is still in the exploratory stage and has not yet been widely used in actual agricultural production. The main bottlenecks include: how to efficiently extract characteristic volatiles related to diseases from complex gas mixtures; how to efficiently process and diagnose volatile group data through intelligent analysis models; and how to transform research results into agricultural application technologies that are easy to operate and cost-controlled. Therefore, combining advanced volatile group detection technology with intelligent data analysis methods to develop an efficient system for tomato root rot detection is of great significance to improving disease management efficiency and realizing precision agriculture. Summary of the invention

[0005] The purpose of the present invention is to address the problems existing in the background technology and to propose a method for detecting tomato root rot based on volatile group information.

[0006] The technical solution of the present invention is a method for detecting tomato root rot based on volatile group information, comprising:

[0007] S1. Construct a tomato root rot detection system based on volatile group information;

[0008] S2. Use a highly sensitive volatile gas sampling device and place it in the root area of ​​the tomato plant to ensure that the device can capture the volatile compounds released by the roots. Determine the sampling time and environmental conditions, perform periodic sampling, and collect the volatile gases from the tomato roots.

[0009] S3, filtering the particulate matter or water vapor that may be mixed in the volatile gas of the tomato roots to obtain a test sample, using the GC-IMS technology to classify and detect the gas components in the test sample, generate <component, concentration> summary data, generate a feature vector, and generate an audit identifier for the <component, concentration> summary data and the feature vector;

[0010] S4. Audit the compliance and completeness of the extracted feature vectors and the summary data of <component, concentration>, input the feature vectors into the pre-trained classification model, obtain the diagnosis results, and calculate the health score. Based on the health score and the diagnosis results, double-check the status of the tomato plants, and output the tomato plant status grade and corresponding strategy based on the root rot grade classification rule table and correction rules;

[0011] S5. The received tomato plant status levels and corresponding strategies are displayed based on the visualization device, and the user manages the tomato plant status levels and corresponding strategies;

[0012] S6. Upload the volatile group data and diagnostic results of each test to the cloud to accumulate historical data and regularly retrain the classification model based on the newly added data.

[0013] Preferably, a tomato root rot detection system based on volatile group information comprises:

[0014] The volatile group information collection module is used to collect volatile gases from tomato roots;

[0015] A data analysis module, used to analyze the volatile components and generate volatile group spectrum data;

[0016] Intelligent diagnosis module, which uses a pre-trained deep learning model to analyze volatile group data and output root rot diagnosis results;

[0017] User interaction module, used to display root rot detection results and prevention and control suggestions;

[0018] The cloud database and model update module is used to store the tomato root rot volatile group feature library and update the detection model.

[0019] Preferably, the process of generating the feature vector is as follows:

[0020] S31. Standardize the concentration data of each volatile substance in the summary data of <ingredient, concentration> and normalize the concentration to the interval [0, 1]:

[0021]

[0022] In the formula, C norm (i) represents the standardized concentration of the i-th volatile compound; C VOCs (i) represents the detected concentration of the i-th volatile compound; C max and C min They represent the maximum and minimum concentrations of all volatiles in the sample, respectively;

[0023] S32. Calculate the relative concentration of each component in the sample:

[0024]

[0025] In the formula, R i represents the relative concentration ratio of the i-th volatile; n represents the number of all volatile species in the volatile group characteristic library; C norm (j) represents the normalized concentration of the jth volatile compound;

[0026] S33, all relative concentration ratios R i Combined into a feature vector R = [R1, R2, ..., R n ].

[0027] Preferably, the audit identifier is generated as follows:

[0028] S4. Select a large prime number p = 2281, and the corresponding p-order irreducible primitive polynomial q = x 2281 +x 715 +1, randomly select an element Satisfy n mod q = 1;

[0029] Where n is a finite field The multiplication group of The order And p is a prime number and satisfies n=2 p -1 is a Mersenne prime; x is an indefinite variable of the polynomial; g≠1;

[0030] S42, select a random number a∈(0,n), calculate the audit code CA=g a mod q, and expose the parameter group {p, g, n, q, CA};

[0031] S43, converting the feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata;

[0032] S44. Calculate identifier generation parameter Idg: Idg = g Bidata modp;

[0033] S45. Randomly select an integer k∈(0,n) and calculate the first-order audit identifier IA1=g k mod q;

[0034] Calculate the second-order audit identifier IA2 = Bidata × k + a × IA1 mod n;

[0035] S46. Generate audit identifier IA=[IA1, IA2].

[0036] Preferably, the audit process for checking the compliance and completeness of the extracted feature vectors and the <component, concentration> summary data is as follows:

[0037] S51, converting the extracted feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata';

[0038] S52. Calculate the first-order audit parameter PI1:

[0039] S53. Calculate the second-order audit parameter PI2:

[0040] Among them, CA is the audit code;

[0041] S54. If PI1=PI2, the extracted {[IA1, IA2], Data, R} passes the audit, that is, the compliance and integrity audit of the extracted feature vector R and the <component, concentration> summary data Data passes.

[0042] Preferably, the diagnostic result is generated as follows:

[0043] S61. A deep fully connected neural network is selected as the classification model. The network structure is as follows:

[0044] S6101, input layer: obtain n-dimensional volatile group feature vector R = [R1, R2, ..., R n ];

[0045] S6102, first hidden layer: extracts high-order nonlinear features, including 2n neurons and ReLU activation function:

[0046] Where n represents the input dimension;

[0047] The second hidden layer: further compresses the feature representation and retains key information, including n neurons and ReLU activation function:

[0048] Regularization: Add Dropout between the second and third hidden layers;

[0049] The third hidden layer: provides low-dimensional feature representation for the classification layer, including Neurons and ReLU activation function:

[0050] in, Represents half of the input dimension, rounded up;

[0051] S6103, output layer: contains 4 neurons, uses Softmax activation function, and outputs the normalized probability of each state:

[0052]

[0053] Where P class,i represents the probability of the i-th category; z class,i represents the i-th category score output by the neural network; C represents the total number of root rot disease status categories;

[0054] S62. Based on the normalized probability of each state, predict the state probability, and output the diagnosis result according to the maximum probability principle:

[0055] P current =[P healthy ,P early ,P mid ,P severe ];

[0056]

[0057] Among them, P current Represents the predicted state probability output by the classification model; P healthy , P early , P mid and P severe They represent the predicted probability of the plant being in a healthy state, the probability of early-stage root rot, the probability of mid-stage root rot, and the probability of severe root rot, respectively; Indicates the output diagnostic results.

[0058] Preferably, the classification model is constructed as follows:

[0059] S71. Obtain historical characteristic data of tomato root rot and use sliding mean to remove sensor noise:

[0060]

[0061] In the formula, Ri ' represents the eigenvalue of the denoised i-th sampling point; k represents the sliding window size; R j represents the j-th dimension original feature vector;

[0062] S72, divide the data set into 70% as training set, 20% as validation set, and 10% as test set;

[0063] S73. Define loss function and accuracy:

[0064]

[0065]

[0066] In the formula, L represents the loss function; m represents the number of batch samples; C represents the number of categories; y i,c Indicates the true label of sample i belonging to category c; P i,c It represents the probability that the model predicts that sample i belongs to category c; TP represents true positive; TN represents true negative; FP represents false positive; FN represents false negative;

[0067] S74. Define an optimization algorithm that combines momentum and adaptive learning rate adjustment to quickly find the global optimal solution:

[0068] The parameter update formula is:

[0069]

[0070] In the formula, θ t represents the parameter after the tth update; θ t-1 represents the parameter after the t-1th update; α represents the learning rate; represents the first-order momentum correction term, which is the exponentially weighted average of the gradient; represents the second-order momentum correction term, which is the exponentially weighted average of the squared gradient; ε represents the smoothing factor, which is used to prevent the denominator from being zero;

[0071] The momentum update formula is:

[0072]

[0073] In the formula, m t represents the first-order momentum, i.e., the gradient moving average; v t represents the second-order momentum, i.e., the average squared gradient; β1 and β2 represent momentum hyperparameters; g t Represents the current gradient;

[0074] Learning rate adjustment: decay the learning rate by 10% every 10 iterations to stabilize convergence;

[0075] S75. Train and obtain a classification model.

[0076] Preferably, the training process of the classification model is as follows:

[0077] S81, initialize the weights and biases of the neural network, and use the Xavier initialization method to ensure gradient stability;

[0078] The initialization formula is:

[0079]

[0080] Where W represents the weight matrix; n in and n out Respectively represent the number of input and output neurons in the current layer; u represents the upper and lower limits of the uniform distribution, which is used to initialize the initial value of the neural network weight W.

[0081] S82, divide the training set into multiple small batches, randomly select a batch of small batch samples from the training set each time, and perform forward and backward propagation;

[0082] S83, the input feature vector passes through each layer of the network, calculates the output probability distribution P, and calculates the loss value L;

[0083] S84. Calculate the gradient of the loss function with respect to each parameter: And use optimization algorithms to adjust parameters;

[0084] S85. After each iteration, use the validation set to evaluate the loss and accuracy to see whether the model performance has improved.

[0085] S86. If the validation set loss does not decrease for 10 consecutive iterations or reaches the set maximum number of training times, stop training, otherwise repeat S81 to S85;

[0086] S87. Test the model with the test set data. If the model meets expectations, save the trained model.

[0087] Preferably, the correction rule is: the range of health score S is used as the basic standard for root rot grade, and the classification results P~ current Provides root rot type information to correct boundary conditions: If P~ current = early, but S ≥ 0.6, directly upgrade the status to "severe disease"; if P ~ current = severe, but S < 0.3, downgrading the status to “moderate disease”;

[0088] Among them, early means early stage of disease; severe means serious disease.

[0089] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0090] The above scheme for detecting tomato root rot based on volatile group information has the following beneficial technical effects:

[0091] The present invention designs a method for detecting tomato root rot based on volatile group information. This scheme not only improves the detection efficiency and accuracy of tomato root rot by integrating multiple advanced technologies and methods, but also has the advantages of real-time monitoring, user-friendliness, data security, and strong adaptability, which has positive significance for promoting the development of modern agricultural technology:

[0092] (1) Improve diagnostic accuracy: By using GC-IMS technology to analyze volatile gases released from tomato roots, combined with advanced deep learning models, different root rot states can be more accurately identified and distinguished. Combined with the generation and verification of audit identifiers, the compliance and integrity of the extracted data can be ensured. The chemical composition-based analysis method is more objective and accurate than traditional visual inspection or empirical judgment.

[0093] (2) Early diagnosis and prevention: The present invention can achieve early diagnosis of tomato root rot, so that farmers can take corresponding prevention and control measures in the early stage of disease development, effectively avoiding further deterioration and spread of the disease;

[0094] (3) Data-driven decision support: Data collected and analyzed through cloud databases can help farmers better understand the development patterns and influencing factors of diseases, providing a scientific basis for formulating more effective management strategies;

[0095] (4) User-friendly interactive interface: The provided mobile application or PC platform allows users to easily obtain test results and prevention and control recommendations, improving the usability and ease of use of the system;

[0096] (5) Continuous learning and optimization: With the accumulation of more data and the continuous training of the classification model, the system's diagnostic capabilities will be further improved and can adapt to the changes in the characteristics of different varieties of tomatoes and multiple diseases;

[0097] (6) Promoting agricultural modernization: The implementation of this plan will help promote the development of traditional agriculture towards intelligence and precision, which is in line with the trend of sustainable development of modern agriculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 A method flow chart of a method for detecting tomato root rot based on volatile group information proposed by the present invention;

[0099] Figure 2 This is a system architecture diagram of a tomato root rot detection system based on volatile group information proposed in the present invention. DETAILED DESCRIPTION

[0100] Embodiment 1, as Figure 1 and Figure 2 As shown in FIG. 1 , the present invention proposes a method for detecting tomato root rot based on volatile group information, such as Figure 1 As shown, the specific implementation steps include the following:

[0101] S1. Construct a tomato root rot detection system based on volatile group information, such as Figure 2 As shown, it includes: a volatile group information collection module, a data analysis module, an intelligent diagnosis module, a user interaction module and a cloud database and model update module;

[0102] The volatile group information collection module is used to collect volatile gases from tomato roots;

[0103] The data analysis module is integrated with GC-IMS (Gas Chromatography-Ion Mobility Spectrometry), which is used to analyze the volatile components and generate volatile component spectrum data;

[0104] The intelligent diagnosis module uses a pre-trained deep learning model to analyze the volatile group data and output the root rot diagnosis results;

[0105] The user interaction module provides a mobile application or PC platform to display root rot detection results and prevention and control suggestions;

[0106] The cloud database and model update module is used to store the tomato root rot volatile group feature library and update the detection model.

[0107] S2. The volatile group information collection module uses a high-sensitivity volatile gas sampling device (including but not limited to a porous adsorption tube or an electronic nose) and arranges it in the root area of ​​the tomato plant to ensure that the equipment can capture the volatile compounds released by the roots, determine the sampling time and environmental conditions (such as sampling during the day when the light is stable), avoid interference from soil moisture or external pollutants, perform periodic sampling, and collect volatile gases from the roots of tomatoes.

[0108] S3, the data analysis module analyzes the collected volatile gases from the tomato roots, and the specific implementation process is as follows:

[0109] S31. Using a physical filtering device to remove particles or water vapor that may be mixed in the volatile gas from the sampled tomato roots, to ensure the purity of the sample, and to obtain a test sample;

[0110] S32. Use GC-IMS technology to classify and detect the gas components in the test sample, separate the volatile components in the sample, detect and identify the components based on the mobility of the compound in the electric field after ionization, and record the gas chromatography retention time (t GC ), IMS migration time (t IMS ) and the corresponding concentration, forming a volatile group characteristic spectrum library, recording each volatile component and its concentration data, and generating <component, concentration> summary data;

[0111] S33, standardize the concentration data of each volatile substance, normalize the concentration to the interval [0, 1], and eliminate the influence of different sample concentration ranges:

[0112]

[0113] In the formula, C norm (i) represents the standardized concentration of the i-th volatile compound; C VOCs (i) represents the detected concentration of the i-th volatile compound; C max and C min They represent the maximum and minimum concentrations of all volatiles in the sample, respectively;

[0114] S34. Calculate the relative concentration of each component in the sample:

[0115]

[0116] In the formula, R i represents the relative concentration ratio of the i-th volatile; n represents the number of all volatile species in the volatile group characteristic library; C norm (j) represents the normalized concentration of the jth volatile compound;

[0117] S35. Set all relative concentration ratios R i Combined into a feature vector R = [R1, R2, ..., R n ];

[0118] S36. Generate an audit identifier for the feature vector and the <component, concentration> summary data. The generation process is as follows:

[0119] S3601, select a large prime number p = 2281, the corresponding p-th order irreducible primitive polynomial q = x 2281 +x 715 +1, randomly select an element Satisfy n mod q = 1;

[0120] Where n is a finite field F 2p The multiplication group of The order And p is a prime number and satisfies n=2 p -1 is a Mersenne prime; x is an indefinite variable of the polynomial; g≠1;

[0121] S3602, select a random number a∈(0,n), calculate the audit code CA=g a mod q, and transmit {p, g, n, q, CA} to the intelligent diagnosis module;

[0122] S3603, converting the feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata;

[0123] S3604, calculate the identifier generation parameter Idg: Idg = g Bidata modp;

[0124] S3605, randomly select an integer k∈(0,n), calculate the first-order audit identifier IA1=g k mod q;

[0125] Calculate the second-order audit identifier IA2 = Bidata × k + a × IA1 mod n;

[0126] S3606, generate audit identifier IA = [IA1, IA2];

[0127] S37, transmitting {[IA1, IA2], Data, R} to the user interaction module and the cloud database and model update module.

[0128] S4, the intelligent diagnosis module extracts {[IA1, IA2], Data, R} from the cloud database and the model update module, builds a classification model adapted to the volatile group characteristics based on the feature vector R, distinguishes the root rot status of tomato plants, and accurately diagnoses tomato root rot. The specific implementation process is as follows:

[0129] S41. Audit the extracted {[IA1, IA2], Data, R} to ensure the compliance and integrity of the extracted feature vector R and the <component, concentration> summary data Data. The audit process is as follows:

[0130] S4101, converting the extracted feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata';

[0131] S4102, calculate the first-order audit parameter PI1: PI1 = g IA2 ;

[0132] S4103, calculate the second-order audit parameter PI2: PI2 = (IA1)Bidata '×(CA) IA1 modq;

[0133] Among them, CA is the audit code;

[0134] S4104. If PI1=PI2, the {[IA1, IA2], Data, R} extracted from the cloud database and the model update module is audited, which means that the compliance and integrity of the extracted feature vector R and the <component, concentration> summary data Data can be guaranteed;

[0135] S42, select the deep fully connected neural network (FCNN) as the core model, the network structure is as follows:

[0136] S4201, input layer: obtain n-dimensional volatile group feature vector R = [R1, R2, ..., R n ];

[0137] S4202, hidden layer:

[0138] (1) The first hidden layer: extracts high-order nonlinear features, including neurons and activation functions:

[0139] Number of neurons: 2n, i.e. twice the input dimension;

[0140] Activation function: ReLU, rectified linear unit;

[0141] (2) The second hidden layer: further compresses the feature representation and retains key information, including neurons and activation functions:

[0142] Number of neurons: n;

[0143] Activation function: ReLU;

[0144] (3) Regularization: Add Dropout between hidden layers to avoid overfitting. In this embodiment, the Dropout probability is set to 0.3, that is, 30% of the neuron connections are randomly discarded;

[0145] (4) The third hidden layer: provides low-dimensional feature representation for the classification layer, including neurons and activation functions:

[0146] Number of neurons: That is, half of the input dimension, rounded up;

[0147] Activation function: ReLU;

[0148] S4203, output layer: contains 4 neurons, uses Softmax activation function, and outputs the normalized probability of each state:

[0149]

[0150] Where P class,i represents the probability of the i-th category; z class,i represents the i-th category score output by the neural network; C represents the total number of root rot status categories, and this embodiment adopts 4 categories, namely healthy, early, medium and severe;

[0151] S43, the feature vector R = [R1, R2, ..., R n ] is input into the pre-trained classification model, the state probability is predicted, and the diagnosis result is output according to the maximum probability principle:

[0152] P current =[P healthy ,P early ,P mid ,P severe ];

[0153]

[0154] Among them, P current Represents the predicted state probability output by the classification model; P healthy , P early , P mid and P severe They represent the predicted probability of the plant being in a healthy state, the probability of early-stage root rot, the probability of mid-stage root rot, and the probability of severe root rot, respectively; Indicates the output diagnostic result;

[0155] S44, calculate confidence = max(P current ), if the confidence Confidence ≥ 0.8, the classification result is directly used Otherwise, execute the following S45;

[0156] S45. Extract the volatile components and their concentration data recorded in the summary data of <Component, Concentration>, combine them with the quantitative assessment of the health score, conduct a secondary verification of the classification results by analyzing the degree of deviation, and then refine the diagnosis and degree of root rot. The specific implementation steps are as follows:

[0157] Calculate the degree of deviation between the current plant and the health status, that is, the health score:

[0158]

[0159] In the formula, S represents the health score. The larger the S is, the more serious the root rot of the plant is; R current (i) represents the i-th item of the current feature; R healthy(i) represents the health feature corresponding to the i-th item of the current feature; n represents the number of all volatile species in the <ingredient, concentration> summary data;

[0160] S46. Based on classification results The health score S, the root rot classification rule table (as shown in Table 1) and the correction rules are used to classify the tomato plant status and output the corresponding strategy;

[0161] Table 1 Root rot classification rules

[0162]

[0163]

[0164] The revised rules are as follows: The range of health score S is used as the basic standard for root rot grade, and the classification results Provides root rot type information to correct some edge cases:

[0165] if But if S ≥ 0.6, the status will be upgraded to "severe disease" directly; if However, if S < 0.3, the status is downgraded to “moderate disease”;

[0166] S47, transmitting the tomato plant root status level and corresponding strategy to the user interaction module and the cloud database and model update module.

[0167] S5. The user interaction module displays the received tomato plant root status level and corresponding strategy based on the visualization device, supports mobile application or PC access, and the user manages the tomato plant root status level and corresponding strategy.

[0168] S6. The cloud database and model update module uploads the volatile group data and diagnostic results of each test to the cloud to accumulate historical data, regularly retrains the classification model based on the new data, optimizes the classification effect, and continuously enriches the volatile group feature library to cover different varieties of tomatoes and multiple disease characteristics, thereby enhancing the applicability of the system.

[0169] Example 2, a method for detecting tomato root rot based on volatile group information proposed by the present invention, compared with Example 1, this embodiment also includes a method for constructing and training a classification model, and its specific implementation steps are as follows:

[0170] S1. Obtain historical characteristic data of tomato root rot and use sliding mean or low-pass filtering to remove sensor noise to ensure accurate characteristic values:

[0171]

[0172] In the formula, Ri ' represents the eigenvalue of the denoised i-th sampling point; k represents the sliding window size; R j represents the j-th dimension original feature vector.

[0173] S2. Divide the data set into 70% as a training set, 20% as a validation set, and 10% as a test set.

[0174] S3. Define loss function and accuracy:

[0175]

[0176]

[0177] Where L represents the loss function; m represents the number of batch samples; C represents the number of categories, which is 4 in this embodiment; y i,c Indicates the true label (one-hot encoding) that sample i belongs to category c; P i,c It represents the probability that the model predicts that sample i belongs to category c; TP represents true positive; TN represents true negative; FP represents false positive; and FN represents false negative.

[0178] S4. Define an optimization algorithm that combines momentum and adaptive learning rate adjustment to quickly find the global optimal solution:

[0179] The parameter update formula is:

[0180]

[0181] In the formula, θ t represents the parameter after the tth update; θ t-1 represents the parameter after the t-1th update; α represents the learning rate, and the initial value is set to 0.001 in this embodiment; represents the first-order momentum correction term, which is the exponentially weighted average of the gradient; represents the second-order momentum correction term, which is the exponentially weighted average of the squared gradient; ε represents the smoothing factor, which is used to prevent the denominator from being zero;

[0182] The momentum update formula is:

[0183]

[0184] In the formula, m t represents the first-order momentum, i.e., the gradient moving average; v t represents the second-order momentum, i.e., the average squared gradient; β1 and β2 represent momentum hyperparameters; g t Represents the current gradient;

[0185] Learning rate adjustment: Decrease the learning rate by 10% every 10 epochs (iterations) to stabilize convergence.

[0186] S5. Train the classification model. The training steps are as follows:

[0187] S51, initialize the weights and biases of the neural network, and use the Xavier initialization method to ensure gradient stability;

[0188] The initialization formula is:

[0189]

[0190] Where W represents the weight matrix; n in and n out Respectively represent the number of input and output neurons in the current layer; u represents the upper and lower limits of the uniform distribution, which is used to initialize the initial value of the neural network weight W.

[0191] S52, dividing the training set into multiple mini-batches. In this embodiment, the batch size is 32. Each time, a batch of samples is randomly selected from the training set for forward and backward propagation.

[0192] S53, the input feature vector passes through each layer of the network, calculates the output probability distribution P, and calculates the loss value L;

[0193] S54. Calculate the gradient of the loss function with respect to each parameter: And use optimization algorithms to adjust parameters;

[0194] S55. After each epoch, use the validation set to evaluate the loss and accuracy to see if the model performance has improved.

[0195] S56. If the validation set loss does not decrease for 10 consecutive epochs or reaches the set maximum number of training times (the maximum number of training times set in this embodiment is 50 epochs), stop training, otherwise repeat S51 to S55;

[0196] S57. Test the model with the test set data. If the model meets expectations, save the trained model and use it as a classification model.

[0197] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A method for detecting tomato root rot based on volatile group information, characterized in that: include S1. Construct a tomato root rot detection system based on volatile group information; S2. Use a highly sensitive volatile gas sampling device and place it in the root area of ​​the tomato plant to ensure that the device can capture the volatile compounds released by the roots. Determine the sampling time and environmental conditions, perform periodic sampling, and collect the volatile gases from the tomato roots. S3, filtering the particulate matter or water vapor that may be mixed in the volatile gas of the tomato roots to obtain a test sample, using the GC-IMS technology to classify and detect the gas components in the test sample, generate <component, concentration> summary data, generate a feature vector, and generate an audit identifier for the <component, concentration> summary data and the feature vector; S4. Audit the compliance and completeness of the extracted feature vectors and the summary data of <component, concentration>, input the feature vectors into the pre-trained classification model, obtain the diagnosis results, and calculate the health score. Based on the health score and the diagnosis results, double-check the status of the tomato plants, and output the tomato plant status grade and corresponding strategy based on the root rot grade classification rule table and correction rules; S5. The received tomato plant status levels and corresponding strategies are displayed based on the visualization device, and the user manages the tomato plant status levels and corresponding strategies; S6. Upload the volatile group data and diagnostic results of each test to the cloud to accumulate historical data and regularly retrain the classification model based on the newly added data.

2. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: A tomato root rot detection system based on volatile group information, comprising: The volatile group information collection module is used to collect volatile gases from tomato roots; A data analysis module, used to analyze the volatile components and generate volatile group spectrum data; Intelligent diagnosis module, which uses a pre-trained deep learning model to analyze volatile group data and output root rot diagnosis results; User interaction module, used to display root rot detection results and prevention and control suggestions; The cloud database and model update module is used to store the tomato root rot volatile group feature library and update the detection model.

3. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: The process of generating feature vectors is as follows: S31. Standardize the concentration data of each volatile substance in the summary data of <ingredient, concentration> and normalize the concentration to the interval [0, 1]: In the formula, C norm (i) represents the standardized concentration of the i-th volatile compound; C VOCs (i) represents the detected concentration of the i-th volatile compound; C max and C min They represent the maximum and minimum concentrations of all volatiles in the sample, respectively; S32. Calculate the relative concentration of each component in the sample: In the formula, R i represents the relative concentration ratio of the i-th volatile; n represents the number of all volatile species in the volatile group characteristic library; C norm (j) represents the normalized concentration of the jth volatile compound; S33, all relative concentration ratios R i Combined into a feature vector R = [R1, R2, ..., R n ].

4. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: The process of generating an audit identifier is as follows: S4. Select a large prime number p = 2281, and the corresponding p-order irreducible primitive polynomial q = x 2281 +x 715 +1, randomly select an element Satisfy n mod q = 1; Where n is a finite field The multiplication group of The order And p is a prime number and satisfies n=2 p -1 is a Mersenne prime; x is an indefinite variable of the polynomial; g≠1; S42, select a random number a∈(0,n), calculate the audit code CA=g a mod q, and expose the parameter group {p, g, n, q, CA}; S43, converting the feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata; S44. Calculate identifier generation parameter Idg: Idg = g Bidata modp; S45. Randomly select an integer k∈(0,n) and calculate the first-order audit identifier IA1=g k mod q; Calculate the second-order audit identifier IA2 = Bidata × k + a × IA1 mod n; S46. Generate audit identifier IA=[IA1, IA2].

5. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: The audit process for compliance and completeness of the extracted feature vectors and <component, concentration> summary data is as follows: S51, converting the extracted feature vector R and the <component, concentration> summary data Data into binary string data, and concatenating them together to obtain the encoded data Bidata'; S52. Calculate the first-order audit parameter PI1: S53. Calculate the second-order audit parameter PI2: Among them, CA is the audit code; S54. If PI1=PI2, the extracted {[IA1, IA2], Data, R} passes the audit, that is, the compliance and integrity audit of the extracted feature vector R and the <component, concentration> summary data Data passes.

6. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: Diagnostics are generated as follows: S61. A deep fully connected neural network is selected as the classification model. The network structure is as follows: S6101, input layer: obtain n-dimensional volatile group feature vector R = [R1, R2, ..., R n ]; S6102, first hidden layer: extracts high-order nonlinear features, including 2n neurons and ReLU activation function: Where n represents the input dimension; The second hidden layer: further compresses the feature representation and retains key information, including n neurons and ReLU activation function: Regularization: Add Dropout between the second and third hidden layers; The third hidden layer: provides low-dimensional feature representation for the classification layer, including Neurons and ReLU activation function: in, Represents half of the input dimension, rounded up; S6103, output layer: contains 4 neurons, uses Softmax activation function, and outputs the normalized probability of each state: Where P class,i represents the probability of the i-th category; z class,i represents the i-th category score output by the neural network; C represents the total number of root rot disease status categories; S62. Based on the normalized probability of each state, predict the state probability, and output the diagnosis result according to the maximum probability principle: P current =[P healthy ,P early ,P mid ,P severe ]; Among them, P current Represents the predicted state probability output by the classification model; P healthy , P early , P mid and P severe They represent the predicted probability of the plant being in a healthy state, the probability of early-stage root rot, the probability of mid-stage root rot, and the probability of severe root rot, respectively; Indicates the output diagnostic results.

7. The method for detecting tomato root rot based on volatile group information according to claim 6, characterized in that: The construction process of the classification model is as follows: S71. Obtain historical characteristic data of tomato root rot and use sliding mean to remove sensor noise: In the formula, R i ' represents the eigenvalue of the denoised i-th sampling point; k represents the sliding window size; R j represents the j-th dimension original feature vector; S72, divide the data set into 70% as training set, 20% as validation set, and 10% as test set; S73. Define loss function and accuracy: In the formula, L represents the loss function; m represents the number of batch samples; C represents the number of categories; y i,c Indicates the true label of sample i belonging to category c; P i,c It represents the probability that the model predicts that sample i belongs to category c; TP represents true positive; TN represents true negative; FP represents false positive; FN represents false negative; S74. Define an optimization algorithm that combines momentum and adaptive learning rate adjustment to quickly find the global optimal solution: The parameter update formula is: In the formula, θ t represents the parameter after the tth update; θ t-1 represents the parameter after the t-1th update; α represents the learning rate; represents the first-order momentum correction term, which is the exponentially weighted average of the gradient; represents the second-order momentum correction term, which is the exponentially weighted average of the squared gradient; ε represents the smoothing factor, which is used to prevent the denominator from being zero; The momentum update formula is: In the formula, m t represents the first-order momentum, i.e., the gradient moving average; v t represents the second-order momentum, i.e., the average squared gradient; β1 and β2 represent momentum hyperparameters; g t Represents the current gradient; Learning rate adjustment: decay the learning rate by 10% every 10 iterations to stabilize convergence; S75. Train and obtain a classification model.

8. The method for detecting tomato root rot based on volatile group information according to claim 6, characterized in that: The training process of the classification model is as follows: S81, initialize the weights and biases of the neural network, and use the Xavier initialization method to ensure gradient stability; The initialization formula is: Where W represents the weight matrix; n in and n out Respectively represent the number of input and output neurons in the current layer; u represents the upper and lower limits of the uniform distribution, which is used to initialize the initial value of the neural network weight W. S82, divide the training set into multiple small batches, randomly select a batch of small batch samples from the training set each time, and perform forward and backward propagation; S83, the input feature vector passes through each layer of the network, calculates the output probability distribution P, and calculates the loss value L; S84. Calculate the gradient of the loss function with respect to each parameter: And use optimization algorithms to adjust parameters; S85. After each iteration, use the validation set to evaluate the loss and accuracy to see whether the model performance has improved. S86. If the validation set loss does not decrease for 10 consecutive iterations or reaches the set maximum number of training times, stop training, otherwise repeat S81 to S85; S87. Test the model with the test set data. If the model meets expectations, save the trained model.

9. The method for detecting tomato root rot based on volatile group information according to claim 1, characterized in that: The revised rule is: the range of health score S is used as the basic standard for root rot grade, and the classification results P~ current Provides root rot type information to correct boundary conditions: If P~ current = early, but S ≥ 0.6, directly upgrade the status to "severe disease"; if P ~ current = severe, but S < 0.3, downgrading the status to "moderate disease"; Among them, early means early stage of disease; severe means serious disease.

Images