Soil environment Cd metal monitoring method based on specific response of root exudates

Abstract

The invention discloses a soil environment Cd metal monitoring method based on specific response of root exudates, which comprises the following steps: in a monitoring period, carrying out root exudates sample collection and soil sample collection on a to-be-monitored area, carrying out root environment perception and dynamic exudates capture based on the collected samples, and generating a root perception data set; a response mode between the soil environment Cd degree and root exudates is established, and an exudates response characteristic fingerprint spectrum is established; coupling the synchronously collected soil environment data to carry out feature importance analysis and regression modeling, quantifying interference rules of different environment factors on a secretion response mode, and generating an environment correction coefficient set; constructing an environmental cadmium analysis model by utilizing the secretion response characteristic fingerprint spectrum and the environmental correction coefficient set; and in the next monitoring period, acquiring real-time area monitoring data of the cadmium sampling points in the area to be monitored, importing the real-time area monitoring data into the environment cadmium analysis model to carry out area cadmium metal situation diagnosis, and carrying out early warning prompt.

Description

Technical Field

[0001] This invention relates to the field of environmental Cd metal monitoring technology, and in particular to a method for monitoring soil Cd metal based on the specific response of root exudates. Background Technology

[0002] In recent years, biomonitoring methods have gained attention as an alternative or supplementary approach, primarily relying on the response of organisms to environmental pollutants to indicate pollution levels. Existing technologies include methods that attempt to reflect soil heavy metal stress using changes in the growth status, leaf morphology, or biomass of specific plant species. However, these phenotypic-based monitoring methods suffer from slow response times, susceptibility to interference from other environmental factors (such as water, nutrients, and pests), and insufficient specificity and sensitivity, making early and accurate diagnosis difficult.

[0003] Plant root exudates are a collective term for various organic compounds released from plant roots into the soil environment, and are one of the most sensitive indicators of plant response to soil environmental stress. Under heavy metal stress, plant root exudation behavior changes significantly, and the types and quantities of specific exudates (such as certain low-molecular-weight organic acids, amino acids, and phenolic substances) exhibit specific and regular changes, which occur long before the appearance of phenotypic symptoms. Therefore, utilizing root exudates—this early and sensitive "chemical language"—to infer and monitor soil Cd pollution has theoretically great potential.

[0004] In summary, a soil environmental Cd metal monitoring method based on the specific response of root exudates is proposed, which enables non-destructive, in-situ, real-time and intelligent monitoring of soil Cd pollution, providing a new technical means for accurate early warning of soil environmental risks and pollution prevention and control. Summary of the Invention

[0005] This invention overcomes the shortcomings of the prior art and provides a method for monitoring Cd metal in the soil environment based on the specific response of root exudates.

[0006] To achieve the above objectives, the first aspect of this invention provides a method for monitoring Cd metal in the soil environment based on the specific response of root exudates, comprising: Within a monitoring cycle, root exudate samples and soil samples are collected from the area to be monitored. Based on the collected samples, root environment perception and dynamic capture of exudates are performed to generate a root perception dataset. Based on the root sensing dataset, the response pattern between soil environmental Cd levels and root exudates was analyzed, stress response pattern features were generated, and exudate response feature fingerprint map was constructed. Based on the fingerprint spectrum of the secretion response characteristics, feature importance analysis and regression modeling are performed by coupling synchronously collected soil environmental data to quantify the interference patterns of different environmental factors on the secretion response pattern and generate a set of environmental correction coefficients. An environmental cadmium analysis model was constructed, and the model was trained using the fingerprint spectrum of the secretion response characteristics and the set of environmental correction coefficients to obtain an environmental cadmium analysis model that meets the expectations. In the next monitoring cycle, real-time regional monitoring data of cadmium sampling points in the area to be monitored are obtained based on the preset sampling scheme, imported into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and early warning prompts are issued.

[0007] In this scheme, within a monitoring cycle, root exudate samples and soil samples are collected from the area to be monitored. Based on the collected samples, root environment sensing and dynamic capture of exudates are performed to generate a root sensing dataset, specifically including: Based on the preset sampling plan, a root exudate collection device is set up in the area to be monitored. Within one monitoring cycle, root exudate collection samples are obtained from each sampling point through the root exudate collection device, and soil samples are obtained from each sampling point. Raman spectrometry was used to detect the root exudate samples collected from each sampling point, and the original Raman spectral signals of the root exudate samples collected from each sampling point were obtained. The original Raman spectrum signal is preprocessed by using an adaptive iterative reweighted penalized least squares method to perform baseline correction on the original spectrum, removing high-frequency noise by using the Savitzky-Golay convolution smoothing algorithm, and then performing vector normalization to obtain the preprocessed Raman spectrum signal. The preprocessed Raman spectral signal was processed by continuous wavelet transform using the Mexican Hat wavelet basis. The characteristic peak positions were extracted by the local extremum detection algorithm. The root exudate species were identified by calling the Raman spectral standard database. The characteristic peak intensities were converted into relative species concentrations to generate root exudate capture information containing time series concentration data. Simultaneously, based on soil samples from each sampling point, inductively coupled plasma mass spectrometry was used to detect cadmium in the soil, and soil environmental characteristic parameters were obtained through soil environmental detection technology to generate root environment perception information. The dynamic time warping algorithm is used to perform spatiotemporal alignment processing on root exudate capture information and root environment perception information, and finally generate root perception dataset.

[0008] In this scheme, the step of analyzing the response pattern between soil environmental Cd levels and root exudates based on the root sensing dataset to generate stress response pattern features specifically includes: A root sensing dataset is obtained, and root exudate features and soil environment features with time-series attributes are extracted from the root sensing dataset. The root exudate features are the main sequence and the soil environment features are the auxiliary sequence. The root sensing feature set is generated by fusing them through time attributes. The root sensing feature set includes root sensing feature sequences of each sampling point in the target area to be monitored. The root sensing feature sequences of each sampling point stored in the root sensing feature set are discretized. The continuous variable is divided into several intervals of the same size using the equal frequency binning method, and the classification is performed using a preset classification threshold to generate a discrete root sensing feature set. The FP-Growth frequent pattern mining algorithm is introduced to analyze the response pattern between soil Cd levels and root exudates. The support of each independent item is obtained by performing a single global scan on the discrete root sensing feature set. The support is compared with a preset minimum support threshold. Independent items with a support greater than the preset minimum support threshold are selected as candidate frequent items and sorted in descending order of support to generate a head table. Then, a second scan is performed on the discrete root system sensing feature set. The items in each sample are sorted according to the head table order, and an FP-tree data structure is recursively constructed. The tree nodes store the item name and count, and the same items are connected through node links. After the FP tree is constructed, all path prefixes containing the target item are extracted from the FP tree, the current item node is removed to form a conditional subtree dataset with the target item as the suffix, and a conditional FP tree is constructed. Based on the constructed conditional FP tree, a bottom-up approach is used to perform reverse path traversal on the target item to obtain all prefix paths and support counts containing the target item, generating a conditional transaction dataset with the target item as a suffix. The generated conditional transaction dataset is then used to construct a new conditional FP tree, and frequent itemset mining is repeatedly performed. The recursion terminates when it is impossible to generate higher-order frequent itemsets. All higher-order frequent itemsets are output. Candidate association rules are generated based on the obtained higher-order frequent itemsets. Valid association rules are then selected based on the support, confidence, and lift of each rule to generate stress response pattern features.

[0009] In this scheme, the construction of the secretion response feature fingerprint map specifically includes: A discrete root system sensing feature set is obtained, and a completely undirected structure graph with connections between all variables is constructed based on the discrete root system sensing feature set as the initial state for causal discovery. The completely undirected structure graph includes all environmental factor variables, root exudate variables, and cadmium concentration variables. Based on the completely undirected structure graph, each variable is sorted from smallest to largest according to its connectivity in the graph. The connectivity is represented by the number of edges owned by each variable node. A connectivity ranking table is generated to define the order of conditional independence tests. For any two variables connected by an edge, condition sets are constructed sequentially from all possible subset combinations of the remaining variables, and the G-test method is used to determine conditional independence. For any set of conditions, calculate the mutual information value between variables in each set of conditions and compare it with a preset threshold. If the value is greater than the preset threshold, the conditions are determined to be independent and the corresponding edge is removed. If the value is less than the preset threshold, the conditions are determined to be independent, the corresponding edge is retained and the dimension of the set of conditions is expanded until all combinations of conditions are traversed and all independent sets of conditions are output. Based on all the independent condition sets in the output, remove all edges that are determined to be conditionally independent from the completely undirected structure graph to form a partially directed acyclic graph. Use the partially directed acyclic graph to extract all unmasked triples, and combine the Meek rule to perform variable causal inference to obtain the causal connection relationship between variables and generate a directed causal network. Obtain stress response pattern features, generate a stress response pattern feature network based on the stress response pattern features, cover the directed causal network graph to mine potential stress response patterns, and obtain the potential stress response pattern feature network. A fingerprint map of secretion response features is constructed based on directed causal network, stress response pattern feature network, and potential stress response pattern feature network.

[0010] In this scheme, based on the exudate response characteristic fingerprint spectrum, feature importance analysis and regression modeling are performed by coupling synchronously acquired soil environmental data to quantify the interference patterns of different environmental factors on exudate response patterns and generate a set of environmental correction coefficients, specifically including: Based on the established correlation response pattern between root exudates and cadmium concentration in the exudate response fingerprint spectrum, response intensity features are extracted, and soil environmental parameters collected simultaneously are acquired, including soil pH, organic matter content, cation exchange capacity and clay content, to generate a coupled dataset. Based on the coupled dataset, the relative importance of each environmental factor to the response pattern of root exudates was analyzed using the random forest algorithm. The response intensity characteristics of root exudates were used as the target variable, and all environmental factors were used as feature variables to construct a random forest. The average reduction in impurity of each environmental factor across all decision trees is calculated using the constructed random forest, which is used as the feature importance score of the environmental factor. Key environmental interference factors are then selected based on the feature importance scores. Based on the key environmental disturbance factors identified through screening, with the response intensity characteristics of root exudates as the dependent variable and the key environmental disturbance factors as the independent variables, the regression coefficients were fitted using the least squares method to obtain a regression equation describing the degree of influence of environmental factors on root exudates. The standardized regression coefficients of each environmental factor are extracted from the regression equation and weighted and fused with the corresponding feature importance scores to calculate the comprehensive interference coefficient of each environmental factor for each root exudate, thereby generating a set of environmental correction coefficients.

[0011] In this scheme, the construction of the environmental cadmium analysis model, which utilizes the secretion response feature fingerprint spectrum and the environmental correction coefficient set for model training to obtain an environmental cadmium analysis model that meets the expectations, specifically includes: Obtain the secretion response feature fingerprint spectrum and the environmental correction coefficient set, import the environmental correction coefficient set into the secretion response feature fingerprint spectrum, and perform environmental factor interference correction by weighting the original response intensity through matrix multiplication operation to generate an environmentally corrected secretion response feature set; The secretion response feature set was aligned and matched with the measured values ​​of soil cadmium concentration, and randomly divided into a training subset and a validation subset according to a preset ratio. The training subset was used for model parameter learning, and the validation subset was used for model performance evaluation and hyperparameter optimization. An environmental cadmium analysis model was constructed based on the support vector regression algorithm. The corrected root exudate response characteristics were used as input variables, and the measured values ​​of soil cadmium concentration were used as target variables. A nonlinear regression model was constructed on the training subset, and the input features were transformed into a high-dimensional space for linear regression analysis through kernel function mapping. The hyperparameters of the environmental cadmium analysis model are optimized by using a grid search strategy combined with cross-validation. By traversing the combination space of penalty coefficient and kernel function parameters, K-fold cross-validation is used to calculate the average mean square error of the model on the training set under each parameter combination. The parameter combination with the smallest average mean square error is selected as the optimal hyperparameters of the model. The environmental cadmium analysis model is retrained on the complete training subset using the optimal hyperparameter combination. The predictive performance of the trained model is evaluated using the validation subset. The coefficient of determination and root mean square error between the predicted and measured values ​​are calculated. When the coefficient of determination is higher than the preset threshold and the root mean square error is lower than the preset tolerance, the model is deemed to meet the expected requirements, and finally, an environmental cadmium analysis model that meets the expectations is obtained.

[0012] In this scheme, the step of acquiring real-time regional monitoring data of cadmium sampling points within the monitoring area based on a preset sampling scheme in the next monitoring cycle, importing the data into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and issuing early warnings specifically includes: In the next monitoring cycle, root exudate samples will be collected from the area to be monitored according to the preset sampling plan. Portable Raman spectroscopy detection equipment will be used to obtain the real-time Raman spectral signals of each sampling point. At the same time, a soil multi-parameter rapid detector will be used to measure soil environmental parameters and generate a real-time regional monitoring dataset. The acquired real-time regional monitoring data is preprocessed and the characteristic peak positions and intensities are extracted. The types of secretions are identified and their relative concentrations are calculated by combining the Raman spectroscopy database. Real-time regional monitoring characteristics are generated by combining real-time soil environmental parameters. The real-time regional monitoring features are input into the environmental cadmium analysis model. The environmental correction coefficient is calculated by similarity to match the current soil environmental parameters to correct the real-time regional monitoring features. The environmental cadmium concentration at each sampling point is predicted by the corrected real-time regional monitoring features to generate environmental cadmium prediction information. Based on the environmental cadmium prediction information, a cadmium concentration distribution map of the area to be monitored is generated using the Kriging spatial interpolation method. A preset cadmium concentration threshold is used to diagnose the regional cadmium pollution situation, generate cadmium pollution situation diagnosis information, and issue early warning prompts.

[0013] A second aspect of the present invention provides a computer-readable storage medium comprising a soil environment Cd metal monitoring method program based on root exudate specific response, wherein when the program is executed by a processor, it implements the steps of the soil environment Cd metal monitoring method based on root exudate specific response as described in any of the preceding claims.

[0014] This invention discloses a method for monitoring Cd metal in the soil environment based on the specific response of root exudates, comprising: collecting root exudate samples and soil samples from the area to be monitored within a monitoring cycle; performing root environment sensing and dynamic capture of exudates based on the collected samples to generate a root sensing dataset; identifying the response pattern between soil Cd levels and root exudates, and constructing an exudate response feature fingerprint spectrum; coupling synchronously collected soil environmental data for feature importance analysis and regression modeling to quantify the interference patterns of different environmental factors on the exudate response pattern, and generating an environmental correction coefficient set; constructing an environmental cadmium analysis model using the exudate response feature fingerprint spectrum and the environmental correction coefficient set; and in the next monitoring cycle, acquiring real-time regional monitoring data of cadmium sampling points within the area to be monitored, importing the data into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and issuing early warnings. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments or examples of the present invention, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0016] Figure 1 A flowchart of the first method for monitoring Cd metal in soil environment based on the specific response of root exudates, provided in an embodiment of the present invention; Figure 2 A second method flowchart of a soil environment Cd metal monitoring method based on root exudate specific response provided in an embodiment of the present invention; Figure 3 A flowchart of the third method for monitoring Cd metal in soil environment based on the specific response of root exudates, provided as an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0018] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0019] Figure 1 A flowchart of the first method for monitoring Cd metal in soil environment based on the specific response of root exudates, provided in an embodiment of the present invention; like Figure 1 As shown, the present invention provides a first method flowchart for a soil environment Cd metal monitoring method based on root exudate-specific response, comprising: S102. Within a monitoring cycle, root exudate samples and soil samples are collected from the area to be monitored. Based on the collected samples, root environment perception and dynamic capture of exudates are performed to generate a root perception dataset. S104, Based on the root sensing dataset, analyze the response pattern between soil environmental Cd level and root exudates, generate stress response pattern features and construct exudate response feature fingerprint map; S106, Based on the secretion response feature fingerprint spectrum, coupled with synchronously collected soil environmental data, feature importance analysis and regression modeling are performed to quantify the interference patterns of different environmental factors on the secretion response pattern and generate an environmental correction coefficient set. S108, Construct an environmental cadmium analysis model, and use the secretion response feature fingerprint spectrum and environmental correction coefficient set to train the model and obtain an environmental cadmium analysis model that meets the expectations. S110, In the next monitoring cycle, real-time regional monitoring data of cadmium sampling points in the area to be monitored are obtained based on the preset sampling scheme, imported into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and early warning is issued.

[0020] Furthermore, in a preferred embodiment of the present invention, the step of collecting root exudate samples and soil samples from the area to be monitored within a monitoring cycle, and performing root environment sensing and dynamic capture of exudates based on the collected samples to generate a root sensing dataset specifically includes: Based on the preset sampling plan, a root exudate collection device is set up in the area to be monitored. Within one monitoring cycle, root exudate collection samples are obtained from each sampling point through the root exudate collection device, and soil samples are obtained from each sampling point. Raman spectrometry was used to detect the root exudate samples collected from each sampling point, and the original Raman spectral signals of the root exudate samples collected from each sampling point were obtained. The original Raman spectrum signal is preprocessed by using an adaptive iterative reweighted penalized least squares method to perform baseline correction on the original spectrum, removing high-frequency noise by using the Savitzky-Golay convolution smoothing algorithm, and then performing vector normalization to obtain the preprocessed Raman spectrum signal. The preprocessed Raman spectral signal was processed by continuous wavelet transform using the Mexican Hat wavelet basis. The characteristic peak positions were extracted by the local extremum detection algorithm. The root exudate species were identified by calling the Raman spectral standard database. The characteristic peak intensities were converted into relative species concentrations to generate root exudate capture information containing time series concentration data. Simultaneously, based on soil samples from each sampling point, inductively coupled plasma mass spectrometry was used to detect cadmium in the soil, and soil environmental characteristic parameters were obtained through soil environmental detection technology to generate root environment perception information. The dynamic time warping algorithm is used to perform spatiotemporal alignment processing on root exudate capture information and root environment perception information, and finally generate root perception dataset.

[0021] It should be noted that, based on a pre-set sampling plan, root exudate collection devices are deployed in the monitoring area. Employing microdialysis or rhizosphere soil solution aspiration principles, root exudate samples can be continuously collected without damaging the plant roots and rhizosphere microenvironment. During a complete monitoring cycle, root exudate solution samples are periodically acquired from each sampling point using this device. Simultaneously, rhizosphere soil samples are collected from the same locations at each sampling point according to corresponding temporal and spatial registration relationships, ensuring the spatiotemporal consistency of exudate and soil samples. After sample acquisition, a high-precision Raman spectrometer is used for rapid detection of the root exudate samples from each sampling point. A 785nm laser light source is used for detection, and raw Raman spectral signals in the 500-1800 cm⁻¹ wavenumber range are obtained by scanning at specific power and integration time.

[0022] Therefore, to extract effective information from the original signal, a series of preprocessing steps are required for the original Raman spectrum. First, an adaptive iterative reweighted penalized least squares method is used for baseline correction, effectively eliminating baseline drift caused by fluorescence background. Then, a Savitzky-Golay convolution smoothing algorithm is used to filter out high-frequency noise and improve the signal-to-noise ratio. Finally, vector normalization is performed to eliminate spectral intensity variations caused by concentration differences, resulting in a high-quality preprocessed Raman spectrum signal. The preprocessed spectral signal is then subjected to continuous wavelet transform using the Mexican Hat wavelet basis. Leveraging the good localization properties of wavelet transform in the time-frequency domain, a local extremum detection algorithm is used to accurately extract characteristic peak positions. The extracted characteristic peak positions are compared with a Raman spectral standard database to identify the specific compounds in the root exudates, such as various organic acids and phenolic substances. Based on the intensity values ​​of the characteristic peaks and a pre-established standard curve, the relative concentration values ​​of various exudates are calculated, ultimately generating root exudate capture information containing time-series concentration data.

[0023] Simultaneously, soil samples collected from each sampling point were analyzed in the laboratory. Inductively coupled plasma mass spectrometry (ICP-MS) was used to determine the total cadmium and available cadmium content in the soil. Meanwhile, conventional soil analysis methods were employed to obtain key environmental characteristic parameters such as pH, organic matter content, and cation exchange capacity, generating comprehensive root environment sensing information. Finally, a dynamic time warping algorithm was used to perform spatiotemporal alignment processing on the root exudate capture information and the root environment sensing information. This algorithm effectively solves the problem of asynchronous data collection times at different sampling points by finding the optimal curved path to align data from different time series, ensuring that all data have a consistent spatiotemporal reference. Ultimately, a root sensing dataset is generated, providing a high-quality data foundation for subsequent response pattern analysis and feature map construction.

[0024] Furthermore, in a preferred embodiment of the present invention, based on the exudate response feature fingerprint spectrum, feature importance analysis and regression modeling are performed by coupling synchronously acquired soil environmental data to quantify the interference patterns of different environmental factors on the exudate response pattern and generate an environmental correction coefficient set, specifically including: Based on the established correlation response pattern between root exudates and cadmium concentration in the exudate response fingerprint spectrum, response intensity features are extracted, and soil environmental parameters collected simultaneously are acquired, including soil pH, organic matter content, cation exchange capacity and clay content, to generate a coupled dataset. Based on the coupled dataset, the relative importance of each environmental factor to the response pattern of root exudates was analyzed using the random forest algorithm. The response intensity characteristics of root exudates were used as the target variable, and all environmental factors were used as feature variables to construct a random forest. The average reduction in impurity of each environmental factor across all decision trees is calculated using the constructed random forest, which is used as the feature importance score of the environmental factor. Key environmental interference factors are then selected based on the feature importance scores. Based on the key environmental disturbance factors identified through screening, with the response intensity characteristics of root exudates as the dependent variable and the key environmental disturbance factors as the independent variables, the regression coefficients were fitted using the least squares method to obtain a regression equation describing the degree of influence of environmental factors on root exudates. The standardized regression coefficients of each environmental factor are extracted from the regression equation and weighted and fused with the corresponding feature importance scores to calculate the comprehensive interference coefficient of each environmental factor for each root exudate, thereby generating a set of environmental correction coefficients.

[0025] It should be noted that the established fingerprint of exudate response features can quantify the response intensity characteristics of the correlation between root exudates and cadmium concentration. This includes key indicators such as peak intensity of specific exudates, concentration change gradient, and response duration, reflecting the feature vector of cadmium stress levels. Simultaneously, soil environmental parameters measured concurrently with exudate sampling are acquired, including key indicators such as soil pH, organic matter content, cation exchange capacity, and clay content. Environmental parameters and exudate response intensity features are precisely matched and fused according to sampling time and spatial location to generate a coupled dataset. Using this coupled dataset, a random forest algorithm is employed to deeply analyze the relative importance of various environmental factors to root exudate response patterns. Using the various response intensity characteristics of root exudates as target variables and all environmental factors as feature variables, a random forest model containing multiple decision trees is constructed. During model training, multiple training subsets are generated using a bootstrap sampling method. A decision tree is constructed for each subset, and the optimal split is performed by randomly selecting a feature subset when splitting nodes. Using a pre-constructed random forest model, the average reduction in impurity resulting from node partitioning across all decision trees for each environmental factor is calculated, and this is used as the feature importance score for that environmental factor. All environmental factors are ranked according to their scores, and key environmental interference factors with importance scores above a preset threshold are selected. Subsequently, based on these key environmental interference factors, a multiple linear regression model is established to quantify the specific impact of environmental factors on exudate response patterns. Using the response intensity characteristics of root exudates as the dependent variable and the key environmental interference factors as independent variables, the least squares method is used to fit the regression coefficients, obtaining a regression equation describing the quantitative relationship between environmental factors and exudate responses. Standardized regression coefficients for each environmental factor are extracted from the regression equation, reflecting the relative contribution of each environmental factor to the exudate response. The standardized regression coefficients are then weighted and fused with the feature importance scores obtained from the random forest, typically using equal weights or differentiated weighting strategies adjusted according to model performance, to calculate the comprehensive interference coefficient for each environmental factor for each root exudate. Finally, an environmental correction coefficient set is generated, containing information such as environmental factor identifiers, exudate type identifiers, comprehensive interference coefficient values, and corresponding confidence intervals. Accurately quantifying the intensity and direction of interference from various environmental factors on the response patterns of exudates under different environmental conditions provides an important basis for the calibration of subsequent environmental cadmium analysis models, ensuring that the models maintain accurate predictive performance under different environmental backgrounds.

[0026] Furthermore, in a preferred embodiment of the present invention, the construction of the environmental cadmium analysis model, which involves training the model using the secretion response feature fingerprint spectrum and the environmental correction coefficient set to obtain an environmental cadmium analysis model that meets the desired outcome, specifically includes: Obtain the secretion response feature fingerprint spectrum and the environmental correction coefficient set, import the environmental correction coefficient set into the secretion response feature fingerprint spectrum, and perform environmental factor interference correction by weighting the original response intensity through matrix multiplication operation to generate an environmentally corrected secretion response feature set; The secretion response feature set was aligned and matched with the measured values ​​of soil cadmium concentration, and randomly divided into a training subset and a validation subset according to a preset ratio. The training subset was used for model parameter learning, and the validation subset was used for model performance evaluation and hyperparameter optimization. An environmental cadmium analysis model was constructed based on the support vector regression algorithm. The corrected root exudate response characteristics were used as input variables, and the measured values ​​of soil cadmium concentration were used as target variables. A nonlinear regression model was constructed on the training subset, and the input features were transformed into a high-dimensional space for linear regression analysis through kernel function mapping. The hyperparameters of the environmental cadmium analysis model are optimized by using a grid search strategy combined with cross-validation. By traversing the combination space of penalty coefficient and kernel function parameters, K-fold cross-validation is used to calculate the average mean square error of the model on the training set under each parameter combination. The parameter combination with the smallest average mean square error is selected as the optimal hyperparameters of the model. The environmental cadmium analysis model is retrained on the complete training subset using the optimal hyperparameter combination. The predictive performance of the trained model is evaluated using the validation subset. The coefficient of determination and root mean square error between the predicted and measured values ​​are calculated. When the coefficient of determination is higher than the preset threshold and the root mean square error is lower than the preset tolerance, the model is deemed to meet the expected requirements, and finally, an environmental cadmium analysis model that meets the expectations is obtained.

[0027] It should be noted that after constructing the fingerprint spectrum of exudate response features and the set of environmental correction coefficients, the set of environmental correction coefficients is imported into the system in matrix form and matrix multiplication is performed with the corresponding root exudate response features in the fingerprint spectrum. Specifically, the original response intensity of each root exudate is weighted and adjusted according to the correction coefficients of its corresponding environmental factors (such as pH, organic matter, etc.), thereby eliminating or reducing signal bias caused by differences in the soil background environment, resulting in the final corrected exudate response feature set. Subsequently, the corrected exudate response feature set is aligned and matched with the measured values ​​of soil cadmium concentration accurately measured by standard laboratory methods (such as ICP-MS) to ensure that the biological characteristic data of each sampling point corresponds one-to-one with the chemical detection results. The entire dataset is randomly divided into a training subset and a validation subset according to a preset ratio. The training subset undertakes the core model learning task and is used to fit the model parameters; the validation subset is used to independently evaluate the model performance during training and provide a basis for subsequent hyperparameter optimization. This is intended to prevent model overfitting and ensure its generalization ability.

[0028] Based on this, Support Vector Regression (SVR) algorithm is adopted as the core framework for constructing the environmental cadmium analysis model. This algorithm is particularly suitable for handling small-sample, nonlinear, high-dimensional data problems. In model construction, the environmentally corrected root exudate response feature matrix is ​​used as the input variable, and the measured soil cadmium concentration is used as the target variable. By performing nonlinear regression analysis on the training subset, the input features are mapped to a higher-dimensional Hilbert space using kernel functions such as radial basis function (RBF), thereby transforming the complex nonlinear relationship in the original feature space into a linear problem in the high-dimensional space, and thus finding the optimal regression hyperplane to establish a complex mapping relationship between the input features and the target value. To obtain the best-performing prediction model, a grid search strategy combined with K-fold cross-validation is used to systematically optimize the key hyperparameters of the SVR model. The grid search automatically traverses the preset combination space of penalty coefficient C and kernel function parameter γ. For each set of candidate parameters, K-fold cross-validation is used to calculate the mean squared error (MSE) of the model prediction on the training set. By comparing the mean MSE under all parameter combinations, the set of parameters that minimizes the error is finally selected as the optimal hyperparameter configuration of the model. Using a determined optimal combination of hyperparameters, the support vector regression model is retrained on the complete training subset to fully utilize all available training data. After training, a reserved validation subset is used to perform a final performance evaluation of the model. The prediction accuracy and reliability of the model are quantified by calculating the coefficient of determination and root mean square error between the predicted and measured values. When the coefficient of determination is higher than a preset threshold and the root mean square error is lower than a preset tolerance, the environmental cadmium analysis model is deemed to meet the expected performance requirements, ultimately resulting in a reliable analysis model that can accurately and stably predict soil cadmium concentration based on root exudate response characteristics.

[0029] Furthermore, in a preferred embodiment of the present invention, the step of acquiring real-time regional monitoring data of cadmium sampling points within the monitoring area based on a preset sampling scheme in the next monitoring cycle, importing the data into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and issuing early warnings specifically includes: In the next monitoring cycle, root exudate samples will be collected from the area to be monitored according to the preset sampling plan. Portable Raman spectroscopy detection equipment will be used to obtain the real-time Raman spectral signals of each sampling point. At the same time, a soil multi-parameter rapid detector will be used to measure soil environmental parameters and generate a real-time regional monitoring dataset. The acquired real-time regional monitoring data is preprocessed and the characteristic peak positions and intensities are extracted. The types of secretions are identified and their relative concentrations are calculated by combining the Raman spectroscopy database. Real-time regional monitoring characteristics are generated by combining real-time soil environmental parameters. The real-time regional monitoring features are input into the environmental cadmium analysis model. The environmental correction coefficient is calculated by similarity to match the current soil environmental parameters to correct the real-time regional monitoring features. The environmental cadmium concentration at each sampling point is predicted by the corrected real-time regional monitoring features to generate environmental cadmium prediction information. Based on the environmental cadmium prediction information, a cadmium concentration distribution map of the area to be monitored is generated using the Kriging spatial interpolation method. A preset cadmium concentration threshold is used to diagnose the regional cadmium pollution situation, generate cadmium pollution situation diagnosis information, and issue early warning prompts.

[0030] It should be noted that in subsequent monitoring cycles, a systematic sampling point network was established in the monitoring area according to a pre-defined gridded sampling scheme. Portable Raman spectroscopy equipment was used to perform in-situ rapid detection of root exudate samples at each sampling point, simultaneously acquiring real-time Raman spectral signals. At the same time, a rapid multi-parameter soil analyzer was used to measure key environmental parameters on-site, such as pH, conductivity, and water content, thereby efficiently generating a real-time regional monitoring dataset containing spectral and environmental factors. This dataset underwent preprocessing, including spectral baseline correction and smoothing / denoising. Feature peak positions and intensity information were extracted through continuous wavelet transform. Combined with a built-in Raman spectral database, exudate types were identified and their relative concentrations calculated. Finally, real-time soil environmental parameters were integrated to form real-time regional monitoring features. These real-time regional monitoring features were input into a pre-trained environmental cadmium analysis model. First, similarity calculations were used to match the most suitable set of environmental correction coefficients under the current soil environmental parameters, accurately correcting the real-time monitoring features to eliminate environmental background interference. Then, the corrected features were used to predict the cadmium concentration at each sampling point using a support vector regression model, generating environmental cadmium prediction information covering the entire region. Based on the predicted cadmium concentration, the spatial distribution surface of cadmium concentration in the entire monitored area is constructed using the Kriging spatial interpolation method. An intuitive pollution distribution map is generated using geographic information system visualization technology. The interpolation results are compared with the preset cadmium concentration safety threshold area by area to automatically identify areas exceeding the standard and calculate the pollution area and intensity. Finally, cadmium pollution situation diagnosis information including pollution level, spatial range and risk level is generated. The platform interface and message push mechanism provide graded early warning prompts, completing the closed-loop intelligent monitoring from data collection to risk decision-making.

[0031] Figure 2 A second method flowchart of a soil environment Cd metal monitoring method based on root exudate specific response provided in an embodiment of the present invention; like Figure 2 As shown, the present invention provides a second method flowchart for monitoring Cd metal in soil environment based on root exudate-specific response, including: S202, Obtain the root sensing dataset, extract root exudate features and soil environment features with time-series attributes from the root sensing dataset, use root exudate features as the main sequence and soil environment features as the auxiliary sequence, fuse them through time attributes to generate a root sensing feature set, the root sensing feature set includes root sensing feature sequences of each sampling point in the target area to be monitored; S204, the root sensing feature sequences of each sampling point stored in the root sensing feature set are discretized, the continuous variable is divided into several intervals of the same size using the equal frequency binning method, and the classification is marked using a preset classification threshold to generate a discrete root sensing feature set. S206 introduces the FP-Growth frequent pattern mining algorithm to analyze the response pattern between soil Cd levels and root exudates. It obtains the support of each independent item by performing a single global scan on the discrete root perception feature set. It then compares the support with a preset minimum support threshold and selects independent items with support greater than the preset minimum support threshold as candidate frequent items. Finally, it sorts the items in descending order of support to generate a head table. S208, then a second scan is performed on the discrete root system sensing feature set. The items in each sample are sorted according to the head table order, and an FP-tree data structure is recursively constructed. The tree nodes store the item name and count, and the same items are connected through node links. After the FP tree is constructed, all path prefixes containing the target item are extracted from the FP tree, the current item node is removed to form a conditional subtree dataset with the target item as the suffix, and a conditional FP tree is constructed. S210: Based on the constructed conditional FP tree, the target item is traversed in reverse from bottom to top to obtain all prefix paths and support counts containing the target item, and a conditional transaction dataset with the target item as the suffix is ​​generated. A new conditional FP tree is constructed using the generated conditional transaction dataset, and frequent itemset mining is repeatedly performed. S212, when it is impossible to generate higher-order frequent itemsets, terminate the recursion, output all higher-order frequent itemsets, generate candidate association rules based on the obtained higher-order frequent itemsets, and filter the effective association rules through the support, confidence and lift of each rule to generate stress response pattern features.

[0032] It should be noted that, firstly, root exudate features and soil environmental features, which possess both temporal attributes, are extracted from the constructed root sensing dataset. Root exudate features are defined as the main sequence because they directly reflect the physiological response of plants; while soil environmental features serve as a secondary sequence, providing environmental background information. The two types of features are fused through timestamp alignment to generate a root sensing feature set. Subsequently, the continuous time series data of each sampling point in the root sensing feature set are discretized. An equal-frequency binning method is used to divide each continuous variable into several intervals, ensuring that the number of samples contained in each interval is basically consistent to maintain the balance of data distribution. Then, each interval is labeled according to a preset classification threshold, for example, dividing concentration values ​​into low, medium, and high categories, thereby transforming continuous temporal features into discrete symbolic sequences and generating a discrete root sensing feature set.

[0033] Building upon this foundation, the FP-Growth frequent pattern mining algorithm is introduced to analyze the response patterns between soil cadmium levels and root exudates. First, a global scan is performed on the discretized feature set to count the frequency (i.e., support) of each independent item (e.g., "high pH," "medium oxalic acid") across the entire dataset. The support of each item is compared to a preset minimum support threshold, filtering out all candidate frequent items that meet the threshold and sorting them in descending order of support to generate a head pointer table. Next, a second scan is performed, sorting the items in each sample according to the head pointer table and recursively constructing an FP-tree data structure. Each node in the FP-tree stores the item name and its count, and nodes with the same name are connected via node links. After the FP-tree is constructed, starting from the bottom of the head pointer table, all prefix paths of each item in the tree are extracted to form a conditional pattern base. After removing the current item node, a conditional subtree dataset with that item as a suffix is ​​constructed, and a conditional FP-tree is built based on this. Based on a conditional FP-tree, a bottom-up approach is used to traverse all paths containing the target item, collecting its prefix paths and corresponding support counts to generate a conditional transaction dataset with the target item as a suffix. This dataset is then used to construct a new conditional FP-tree, and a frequent itemset mining process is recursively executed to continuously expand the itemset size. When no higher-order frequent itemsets satisfying the minimum support threshold can be generated, the recursive process terminates, and all higher-order frequent itemsets are output. Finally, candidate association rules are generated based on all the obtained higher-order frequent itemsets. The support, confidence, and lift indices of each rule are calculated, and effective rules with both high frequency and strong correlation are selected. These rules reveal the characteristic response patterns of root exudates under specific cadmium pollution levels, ultimately forming a pattern feature set characterizing the stress response mechanism of the soil-plant system.

[0034] Figure 3A flowchart of the third method for monitoring Cd metal in soil environment based on the specific response of root exudates, provided as an embodiment of the present invention; like Figure 3 As shown, this invention provides a third method flowchart for monitoring Cd metal in soil environment based on root exudate-specific response, including: S302, Obtain a discrete root system sensing feature set, and construct a completely undirected structure graph based on the discrete root system sensing feature set as the initial state for causal discovery, wherein the completely undirected structure graph contains all environmental factor variables, root exudate variables and cadmium concentration variables. S304, Based on the completely undirected structure graph, sort the variables in the graph from smallest to largest according to their connectivity, where connectivity is represented by the number of edges each variable node has, and generate a connectivity ranking table to define the order of conditional independence tests; for any two variables connected by an edge, construct condition sets in order from all possible subset combinations of the remaining variables and use the G-test method to determine conditional independence. S306. For any set of conditions, calculate the mutual information value between variables in each set of conditions and compare it with a preset threshold. If the value is greater than the preset threshold, the conditions are determined to be independent and the corresponding edge is removed. If the value is less than the preset threshold, the conditions are determined to be independent, the corresponding edge is retained and the dimension of the set of conditions is expanded until all combinations of conditions are traversed and all independent sets of conditions are output. S308, based on all the independent condition sets output, remove all edges that are determined to be conditionally independent in the completely undirected structure graph to form a partially directed acyclic graph. Use the partially directed acyclic graph to extract all unmasked triples, combine the Meek rule to perform variable causal inference to obtain the causal connection relationship between variables and generate a directed causal network. S310, Obtain stress response pattern features, generate a stress response pattern feature network based on the stress response pattern features, cover the directed causal network graph to mine potential stress response patterns, and obtain the potential stress response pattern feature network. S312, construct a secretion response feature fingerprint map based on the directed causal network, the stress response pattern feature network, and the potential stress response pattern feature network.

[0035] It should be noted that after completing the association rule mining, the response pattern characteristics between root exudates and soil cadmium concentration are analyzed through association rule analysis. The response pattern characteristics obtained at this time are strong correlation responses, that is, direct characteristics that can directly indicate the change law of cadmium concentration and root exudates. However, there are still implicit correlation characteristics. Therefore, implicit correlation pattern mining is achieved through causal inference.

[0036] First, a complete undirected graph containing all variables is constructed based on the discrete root sensing feature set as the initial framework for causal discovery. This graph includes all environmental factor variables (such as pH value and organic matter content), root exudate variables (such as oxalic acid concentration and malic acid concentration), and cadmium concentration variables, and establishes connecting edges between each pair of variables to form an initial fully connected network structure. Subsequently, based on this completely undirected graph, the connectivity degree (i.e., the number of edges connected to that node) of each variable is calculated and sorted according to the connectivity degree values ​​from smallest to largest, generating a connectivity degree ranking table to represent the execution order of conditional independence tests, prioritizing the testing of variable pairs with lower connectivity degrees to improve computational efficiency. For any two variables directly connected by edges in the graph, a condition set is systematically constructed from all possible subset combinations of the remaining variables, and the G-test method is used to determine conditional independence. Specifically, for variables X and Y, given a condition set Z, the conditional mutual information value is calculated and transformed into a G-statistic. This G-statistic is then compared with a preset significance threshold (usually set to p=0.01): if the p-value is greater than the threshold, X and Y are determined to be independent given Z, and the corresponding edge is removed; if the p-value is less than or equal to the threshold, the edge is retained, and the dimension of the condition set is expanded (e.g., from a univariate to a multivariate condition set) until all possible combinations of condition sets are traversed. Finally, all independent condition sets that pass the test are output. Based on the test results of all independent condition sets, all edges determined to be conditionally independent in the completely undirected graph are removed, forming the skeleton structure of a partially directed acyclic graph (PDAG).

[0037] Subsequently, all unmasked triples (i.e., structures where X, Y, and Z are not adjacent) in the framework are identified. The causal direction is determined by examining the independence of X and Z under given Y and other variables: if conditional independence exists, it is identified as a collision structure (X→Y←Z). Based on this, the Meek rule series is applied to propagate and deduce the causal direction, gradually determining all possible causal directions, ultimately generating a complete directed causal network. The obtained stress response pattern features (i.e., strong association rules) are mapped onto the directed causal network to construct a stress response pattern feature network. Through latent pattern mining in this network, connections that do not reach the strong association rule threshold but have stable causal relationships in the causal network are identified, forming a potential stress response pattern feature network. Finally, the directed causal network, the stress response pattern feature network, and the potential stress response pattern feature network are integrated. By calculating parameters such as node centrality index, edge weight, and causal path strength, a feature fingerprint map that comprehensively characterizes the soil-plant system response mechanism under Cd stress is constructed. This reveals the causal mechanism and potential response patterns between variables, providing a complete characteristic representation system for understanding biological responses under cadmium pollution stress.

[0038] In another aspect, the present invention provides a computer-readable storage medium comprising a soil environment Cd metal monitoring method program based on root exudate specific response. When the program is executed by a processor, it implements the steps of the soil environment Cd metal monitoring method based on root exudate specific response as described in any of the preceding claims.

[0039] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0040] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0041] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0042] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0043] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0044] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for monitoring Cd metal in soil environment based on the specific response of root exudates, characterized in that, include: Within a monitoring cycle, root exudate samples and soil samples are collected from the area to be monitored. Based on the collected samples, root environment perception and dynamic capture of exudates are performed to generate a root perception dataset. Based on the root sensing dataset, the response pattern between soil environmental Cd levels and root exudates was analyzed, stress response pattern features were generated, and exudate response feature fingerprint map was constructed. Based on the fingerprint spectrum of the secretion response characteristics, feature importance analysis and regression modeling are performed by coupling synchronously collected soil environmental data to quantify the interference patterns of different environmental factors on the secretion response pattern and generate a set of environmental correction coefficients. An environmental cadmium analysis model was constructed, and the model was trained using the fingerprint spectrum of the secretion response characteristics and the set of environmental correction coefficients to obtain an environmental cadmium analysis model that meets the expectations. In the next monitoring cycle, real-time regional monitoring data of cadmium sampling points in the area to be monitored are obtained based on the preset sampling scheme, imported into the environmental cadmium analysis model for regional cadmium metal status diagnosis, and early warning prompts are issued.

2. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, Within a monitoring cycle, root exudate and soil samples are collected from the area to be monitored. Based on the collected samples, root environment sensing and dynamic capture of exudates are performed to generate a root sensing dataset, specifically including: Based on the preset sampling plan, a root exudate collection device is set up in the area to be monitored. Within one monitoring cycle, root exudate collection samples are obtained from each sampling point through the root exudate collection device, and soil samples are obtained from each sampling point. Raman spectrometry was used to detect the root exudate samples collected from each sampling point, and the original Raman spectral signals of the root exudate samples collected from each sampling point were obtained. The original Raman spectrum signal is preprocessed by using an adaptive iterative reweighted penalized least squares method to perform baseline correction on the original spectrum, removing high-frequency noise by using the Savitzky-Golay convolution smoothing algorithm, and then performing vector normalization to obtain the preprocessed Raman spectrum signal. The preprocessed Raman spectral signal was processed by continuous wavelet transform using the Mexican Hat wavelet basis. The characteristic peak positions were extracted by the local extremum detection algorithm. The root exudate species were identified by calling the Raman spectral standard database. The characteristic peak intensities were converted into relative species concentrations to generate root exudate capture information containing time series concentration data. Simultaneously, based on soil samples from each sampling point, inductively coupled plasma mass spectrometry was used to detect cadmium in the soil, and soil environmental characteristic parameters were obtained through soil environmental detection technology to generate root environment perception information. The dynamic time warping algorithm is used to perform spatiotemporal alignment processing on root exudate capture information and root environment perception information, and finally generate root perception dataset.

3. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, The analysis of the response pattern between soil environmental Cd levels and root exudates based on the root sensing dataset, generating stress response pattern features, specifically includes: A root sensing dataset is obtained, and root exudate features and soil environment features with time-series attributes are extracted from the root sensing dataset. The root exudate features are the main sequence and the soil environment features are the auxiliary sequence. The root sensing feature set is generated by fusing them through time attributes. The root sensing feature set includes root sensing feature sequences of each sampling point in the target area to be monitored. The root sensing feature sequences of each sampling point stored in the root sensing feature set are discretized. The continuous variable is divided into several intervals of the same size using the equal frequency binning method, and the classification is performed using a preset classification threshold to generate a discrete root sensing feature set. The FP-Growth frequent pattern mining algorithm is introduced to analyze the response pattern between soil Cd levels and root exudates. The support of each independent item is obtained by performing a single global scan on the discrete root sensing feature set. The support is compared with a preset minimum support threshold. Independent items with a support greater than the preset minimum support threshold are selected as candidate frequent items and sorted in descending order of support to generate a head table. Then, a second scan is performed on the discrete root system sensing feature set. The items in each sample are sorted according to the head table order, and an FP-tree data structure is recursively constructed. The tree nodes store the item name and count, and the same items are connected through node links. After the FP tree is constructed, all path prefixes containing the target item are extracted from the FP tree, the current item node is removed to form a conditional subtree dataset with the target item as the suffix, and a conditional FP tree is constructed. Based on the constructed conditional FP tree, a bottom-up approach is used to perform reverse path traversal on the target item to obtain all prefix paths and support counts containing the target item, generating a conditional transaction dataset with the target item as a suffix. The generated conditional transaction dataset is then used to construct a new conditional FP tree, and frequent itemset mining is repeatedly performed. The recursion terminates when it is impossible to generate higher-order frequent itemsets. All higher-order frequent itemsets are output. Candidate association rules are generated based on the obtained higher-order frequent itemsets. Valid association rules are then selected based on the support, confidence, and lift of each rule to generate stress response pattern features.

4. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, The construction of the secretion response feature fingerprint map specifically includes: A discrete root system sensing feature set is obtained, and a completely undirected structure graph with connections between all variables is constructed based on the discrete root system sensing feature set as the initial state for causal discovery. The completely undirected structure graph includes all environmental factor variables, root exudate variables, and cadmium concentration variables. Based on the completely undirected structure graph, each variable is sorted from smallest to largest according to its connectivity in the graph. The connectivity is represented by the number of edges owned by each variable node. A connectivity ranking table is generated to define the order of conditional independence tests. For any two variables connected by an edge, condition sets are constructed sequentially from all possible subset combinations of the remaining variables, and the G-test method is used to determine conditional independence. For any set of conditions, calculate the mutual information value between variables in each set of conditions and compare it with a preset threshold. If the value is greater than the preset threshold, the conditions are determined to be independent and the corresponding edge is removed. If the value is less than the preset threshold, the conditions are determined to be independent, the corresponding edge is retained and the dimension of the set of conditions is expanded until all combinations of conditions are traversed and all independent sets of conditions are output. Based on all the independent condition sets in the output, remove all edges that are determined to be conditionally independent from the completely undirected structure graph to form a partially directed acyclic graph. Use the partially directed acyclic graph to extract all unmasked triples, and combine the Meek rule to perform variable causal inference to obtain the causal connection relationship between variables and generate a directed causal network. Obtain stress response pattern features, generate a stress response pattern feature network based on the stress response pattern features, cover the directed causal network graph to mine potential stress response patterns, and obtain the potential stress response pattern feature network. A fingerprint map of secretion response features is constructed based on directed causal network, stress response pattern feature network, and potential stress response pattern feature network.

5. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, Based on the exudate response characteristic fingerprint spectrum, feature importance analysis and regression modeling are performed by coupling synchronously acquired soil environmental data to quantify the interference patterns of different environmental factors on exudate response patterns and generate a set of environmental correction coefficients, specifically including: Based on the established correlation response pattern between root exudates and cadmium concentration in the exudate response fingerprint spectrum, response intensity features are extracted, and soil environmental parameters collected simultaneously are acquired, including soil pH, organic matter content, cation exchange capacity and clay content, to generate a coupled dataset. Based on the coupled dataset, the relative importance of each environmental factor to the response pattern of root exudates was analyzed using the random forest algorithm. The response intensity characteristics of root exudates were used as the target variable, and all environmental factors were used as feature variables to construct a random forest. The average reduction in impurity of each environmental factor across all decision trees is calculated using the constructed random forest, which is used as the feature importance score of the environmental factor. Key environmental interference factors are then selected based on the feature importance scores. Based on the key environmental disturbance factors identified through screening, with the response intensity characteristics of root exudates as the dependent variable and the key environmental disturbance factors as the independent variables, the regression coefficients were fitted using the least squares method to obtain a regression equation describing the degree of influence of environmental factors on root exudates. The standardized regression coefficients of each environmental factor are extracted from the regression equation and weighted and fused with the corresponding feature importance scores to calculate the comprehensive interference coefficient of each environmental factor for each root exudate, thereby generating a set of environmental correction coefficients.

6. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, The construction of the environmental cadmium analysis model involves training the model using the exudate response feature fingerprint spectrum and the environmental correction coefficient set to obtain an environmental cadmium analysis model that meets the desired outcome. Specifically, this includes: Obtain the secretion response feature fingerprint spectrum and the environmental correction coefficient set, import the environmental correction coefficient set into the secretion response feature fingerprint spectrum, and perform environmental factor interference correction by weighting the original response intensity through matrix multiplication operation to generate an environmentally corrected secretion response feature set; The secretion response feature set was aligned and matched with the measured values ​​of soil cadmium concentration, and randomly divided into a training subset and a validation subset according to a preset ratio. The training subset was used for model parameter learning, and the validation subset was used for model performance evaluation and hyperparameter optimization. An environmental cadmium analysis model was constructed based on the support vector regression algorithm. The corrected root exudate response characteristics were used as input variables, and the measured values ​​of soil cadmium concentration were used as target variables. A nonlinear regression model was constructed on the training subset, and the input features were transformed into a high-dimensional space for linear regression analysis through kernel function mapping. The hyperparameters of the environmental cadmium analysis model are optimized by using a grid search strategy combined with cross-validation. By traversing the combination space of penalty coefficient and kernel function parameters, K-fold cross-validation is used to calculate the average mean square error of the model on the training set under each parameter combination. The parameter combination with the smallest average mean square error is selected as the optimal hyperparameters of the model. The environmental cadmium analysis model is retrained on the complete training subset using the optimal hyperparameter combination. The predictive performance of the trained model is evaluated using the validation subset. The coefficient of determination and root mean square error between the predicted and measured values ​​are calculated. When the coefficient of determination is higher than the preset threshold and the root mean square error is lower than the preset tolerance, the model is deemed to meet the expected requirements, and finally, an environmental cadmium analysis model that meets the expectations is obtained.

7. The method for monitoring Cd metal in soil environment based on the specific response of root exudates according to claim 1, characterized in that, In the next monitoring cycle, real-time regional monitoring data of cadmium sampling points within the monitored area are obtained based on a preset sampling scheme. This data is then imported into the environmental cadmium analysis model for regional cadmium metal status diagnosis and early warning alerts are issued. Specifically, this includes: In the next monitoring cycle, root exudate samples will be collected from the area to be monitored according to the preset sampling plan. Portable Raman spectroscopy detection equipment will be used to obtain the real-time Raman spectral signals of each sampling point. At the same time, a soil multi-parameter rapid detector will be used to measure soil environmental parameters and generate a real-time regional monitoring dataset. The acquired real-time regional monitoring data is preprocessed and the characteristic peak positions and intensities are extracted. The types of secretions are identified and their relative concentrations are calculated by combining the Raman spectroscopy database. Real-time regional monitoring characteristics are generated by combining real-time soil environmental parameters. The real-time regional monitoring features are input into the environmental cadmium analysis model. The environmental correction coefficient is calculated by similarity to match the current soil environmental parameters to correct the real-time regional monitoring features. The environmental cadmium concentration at each sampling point is predicted by the corrected real-time regional monitoring features to generate environmental cadmium prediction information. Based on the environmental cadmium prediction information, a cadmium concentration distribution map of the area to be monitored is generated using the Kriging spatial interpolation method. A preset cadmium concentration threshold is used to diagnose the regional cadmium pollution situation, generate cadmium pollution situation diagnosis information, and issue early warning prompts.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a soil environment Cd metal monitoring method program based on root exudate specific response. When the program is executed by a processor, it implements the steps of the soil environment Cd metal monitoring method based on root exudate specific response as described in any one of claims 1 to 7.

Images