A method for detecting and analyzing soil in contaminated sites

Through recursive random field model and multimodal coupled analysis, the problems of low accuracy and prediction instability of traditional soil detection methods are solved, and high-precision pollutant concentration prediction and spatial distribution analysis are achieved, supporting scientific environmental governance of polluted sites.

CN118897072BActive Publication Date: 2024-12-06赤峰市生态环境监控中心
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411387074.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-30
Publication Date
2024-12-06
Estimated Expiration
2044-09-30

AI Technical Summary

Technical Problem

Traditional soil detection methods have low accuracy in identifying pollutants, are prone to missed detection or misjudgment, lack effective high-order dependency modeling capabilities, resulting in unstable prediction performance, and it is difficult to accurately capture the spatial distribution and concentration changes of pollutants. Provide sufficient scientific basis for environmental governance of polluted sites.

Method used

Using recursive random field model and multimodal coupling analysis method, the recursive random field model is constructed by obtaining the multimodal data set of soil samples, calculating the coupling matrix, and dynamically adjusting it for global optimization to predict the concentration and distribution of pollutants. Generate a pollutant concentration distribution map, conduct risk assessment and remediation plans.

Benefits of technology

It improves the accuracy and recognition accuracy of pollutant concentration prediction, enhances the flexibility and robustness of the model, can maintain efficient prediction performance in complex environments, provides detailed pollutant spatial distribution maps, and supports scientific environmental governance decisions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118897072B_ABST
    Figure CN118897072B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of physical or chemical analysis technology, and in particular to a method for detecting and analyzing soil at a contaminated site. The content includes: obtaining the reflection data of the physical and chemical properties of soil samples at the contaminated site to obtain a multimodal data set; constructing a recursive random field model based on the multimodal data set; calculating the coupling matrix through a multimodal coupling analysis method, globally optimizing the multimodal coupling analysis process, and dynamically adjusting the coupling matrix to obtain an updated coupling matrix; based on the updated coupling matrix, obtaining the predicted concentration value of the pollutant, and detecting and analyzing the soil at the contaminated site. It solves the problem that traditional soil methods are prone to missed detection or misjudgment of pollutants, lack of effective high-order dependency modeling capabilities, resulting in unstable prediction performance, poor adaptability, difficulty in accurately capturing the spatial distribution and concentration changes of pollutants, and failure to provide sufficient scientific basis for environmental governance of contaminated sites.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of physical or chemical analysis technology, and in particular to a method for detecting and analyzing soil in a contaminated site. Background Art

[0002] Globally, with the acceleration of industrialization and the widespread use of agricultural chemicals, soil pollution has become increasingly serious and has become an environmental problem that needs to be solved urgently. As an important part of the ecosystem, soil directly affects the growth of plants, the habitat of animals and human health. Once the soil is polluted, especially by harmful substances such as heavy metals and organic pollutants, it will not only lead to the decline of agricultural product quality and the deterioration of the ecological environment, but may also pose a serious threat to human health through the food chain or direct contact. Therefore, accurate and rapid detection and analysis of soil in contaminated sites has become one of the key tasks in environmental protection and land remediation.

[0003] In recent years, with the advancement of environmental monitoring technology, soil testing methods have also made certain progress. However, with the increase in the types of pollutants and the increasing complexity of their distribution, traditional soil testing methods have been unable to meet the growing monitoring needs. Modern soil testing not only requires the ability to detect a variety of pollutants, but also requires high detection accuracy and speed. In addition, contaminated sites often have heterogeneous and complex spatial distribution characteristics, which requires detection technology to capture subtle changes in various pollutants in soil samples and conduct detailed analysis of their spatial distribution.

[0004] In summary, traditional soil detection methods have the following technical problems: low accuracy in identifying pollutants, prone to missed detection or misjudgment, lack of effective high-order dependency modeling capabilities, resulting in unstable prediction performance and poor adaptability, making it difficult to accurately capture the spatial distribution and concentration changes of pollutants, and unable to provide sufficient scientific basis for environmental governance of contaminated sites. Summary of the invention

[0005] The present invention provides a soil detection and analysis method for contaminated sites to solve the technical problems that traditional soil detection methods have low accuracy in identifying pollutants, are prone to missed detection or misjudgment, lack effective high-order dependency modeling capabilities, resulting in unstable prediction performance and poor adaptability, and are difficult to accurately capture the spatial distribution and concentration changes of pollutants, and cannot provide sufficient scientific basis for environmental governance of contaminated sites.

[0006] A method for detecting and analyzing soil in a contaminated site of the present invention specifically includes the following technical solutions:

[0007] A method for detecting and analyzing soil in a contaminated site comprises the following steps:

[0008] S1. Obtain the data reflecting the physical and chemical properties of soil samples from contaminated sites to obtain a multimodal data set; based on the multimodal data set, construct a recursive random field model;

[0009] S2. Calculate the coupling matrix through the multimodal coupling analysis method, globally optimize the multimodal coupling analysis process, and dynamically adjust the coupling matrix to obtain an updated coupling matrix; obtain the predicted concentration value of the pollutant based on the updated coupling matrix; and detect and analyze the soil of the contaminated site based on the predicted concentration value of the pollutant.

[0010] Preferably, the S1 specifically includes:

[0011] In the process of constructing the recursive random field model, the recursive random field model is initialized and the probability density function of the recursive random field model is obtained.

[0012] Preferably, the S1 specifically includes:

[0013] The coupling relationship between soil samples is gradually adjusted through recursive optimization. In each recursion, the recursive random field model optimizes the coupling coefficient between samples by minimizing the objective function to obtain the optimized coupling coefficient between samples.

[0014] Preferably, the S2 specifically includes:

[0015] Based on the optimized coupling coefficient between samples, the characteristic vectors of soil samples under different modes are coupled and analyzed by multimodal coupling analysis method to obtain the coupling matrix. The specific calculation formula of the coupling matrix is ​​as follows:

[0016] in, is at the time step The resulting coupling matrix; is modal and The coupling coefficient between is the number of modal types; is the coupling coefficient between the optimized samples; Representing modality Datasets; Representing modality Datasets; is the transformation matrix between modes; represents transpose; is the bias term; are the adjustment weights of the recursive random field model; is the pollutant concentration; is the probability density function of the recursive random field model.

[0017] Preferably, the S2 specifically includes:

[0018] In the process of global optimization of the multi-modal coupling analysis process, a recursive optimization objective function is defined and minimized.

[0019] Preferably, the S2 specifically includes:

[0020] An adaptive coupling mechanism is introduced to dynamically adjust the coupling matrix to obtain an updated coupling matrix.

[0021] Preferably, the S2 specifically includes:

[0022] Based on the updated coupling matrix, the concentration and distribution of pollutants are predicted through the regression model to obtain the predicted concentration value of the pollutant.

[0023] Preferably, the S2 specifically includes:

[0024] Generate a pollutant concentration distribution map based on the predicted pollutant concentration values ​​and the actual geographical location information of the soil samples obtained; conduct a risk assessment of pollutant concentrations in different areas based on the pollutant concentration distribution map to obtain pollution risk assessment results; and formulate a soil remediation plan based on the pollution risk assessment results.

[0025] The beneficial effects of the technical solution of the present invention are:

[0026] 1. By processing multimodal data sets with a recursive random field model, it is possible to capture the complex high-order dependencies between different modal data. Combining multimodal coupling analysis with recursive optimization technology ensures high-precision predictions of pollutant concentrations. This improves the accuracy of pollutant identification and reduces the deviations that may be introduced by traditional single-modal soil detection methods.

[0027] 2. An adaptive coupling mechanism is introduced to dynamically adjust the coupling matrix during the recursive optimization process so that it can adapt to the changing modal relationships during the recursive process, ensuring that the recursive random field model can still maintain efficient prediction performance when faced with complex and changeable soil samples; compared with traditional static models, the method of the present invention is more flexible and robust, and can maintain prediction accuracy under changing environmental conditions.

[0028] 3. The coupling relationship between soil samples is gradually adjusted through recursive optimization, and the entire multimodal coupling analysis process is globally optimized; recursive optimization enables the recursive random field model to capture the complex interaction relationship in multimodal data at a deeper level, thereby improving the overall prediction performance; global optimization further ensures the accuracy and stability of the recursive random field model, especially when dealing with multimodal and large-scale data, it has significant advantages.

[0029] 4. Based on the updated coupling matrix, the data in the multimodal data set are mapped to the pollutant concentration space through the regression model, which can accurately predict the concentration and distribution of pollutants in the soil; interpolation analysis is performed in combination with the actual geographical location information of the soil samples to generate a detailed pollutant concentration distribution map, which helps to intuitively display the spatial distribution of pollutants in the site. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The present invention is a flow chart of a method for detecting and analyzing soil in a contaminated site. DETAILED DESCRIPTION

[0031] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0033] The specific scheme of the soil detection and analysis method for contaminated sites provided by the present invention is described in detail below with reference to the accompanying drawings.

[0034] See attached Figure 1 , which shows a flow chart of a method for detecting and analyzing soil in a contaminated site provided by an embodiment of the present invention, the method comprising the following steps:

[0035] S1. Obtain the data reflecting the physical and chemical properties of soil samples from contaminated sites to obtain a multimodal data set; based on the multimodal data set, construct a recursive random field model;

[0036] Soil samples from contaminated sites are tested using a multimodal sensor array to obtain data on multiple physical and chemical properties of soil samples, including spectral reflectance, pH value, conductivity, water content, etc. The collected reflection data is stored as a data set in the form of modes to obtain a multimodal data set, where each mode represents a physical or chemical property.

[0037] The dataset for each modality is represented as a matrix ,in Indicates the modality type, and the feature dimension is The formula for constructing the dataset of each modality is as follows:

[0038]

[0039] in, Representing modality Next The feature vector of each soil sample is Contains the soil sample in the modal Data on all feature dimensions in; represents transpose; Indicates that in modal Next Soil samples in characteristic dimensions data.

[0040] By constructing a recursive random field model, the complex dependencies between the data of each modality in a multimodal dataset are processed; the purpose of constructing a recursive random field model is to capture the high-order dependencies between different modal data and model the dependencies through a recursive structure. The construction of a recursive random field model specifically includes the following steps:

[0041] The first step is to initialize the recursive random field model: initialize the corresponding random field on each modal data set to represent the relationship between the feature vectors of each soil sample. The formula of the recursive random field model is as follows:

[0042]

[0043] in, Represents the probability density function of the recursive random field model, used to represent the mode Dataset and the target variable the interrelationships between (pollutant concentrations); Contains modal The characteristic vectors of all soil samples under is modal Next Soil samples and Coupling coefficient between soil samples; It is the measurement mode The characteristic vector of the soil sample and The function of the similarity between them can use Gaussian kernel or polynomial kernel function; is modal Next The feature vector of the soil sample With Contaminant concentration labels for soil samples The mapping function between is the number of modal types, i.e., the total number of physical or chemical properties analyzed in the soil sample. Through the recursive structure, the complex interactions between the characteristic vectors of the soil samples are captured and provide a basis for the subsequent coupling analysis.

[0044] The second step is recursive optimization: the coupling relationship between soil samples is gradually adjusted through recursive optimization to maximize the accuracy of the recursive random field model; in each recursion, the recursive random field model optimizes the coupling coefficient between samples by minimizing the following objective function:

[0045]

[0046] in, is the objective function used to evaluate the Soil samples and The coupling coefficient between soil samples The quality of the algorithm and guide the recursive optimization process; is a regularization parameter used to prevent overfitting of the recursive random field model; yes Regularization term, used to control the Soil samples and The complexity of the coupling coefficient between soil samples is minimized by , and the optimized coupling coefficient between samples is obtained; by optimizing the coupling relationship between samples, the recursive random field model can accurately capture the complex dependencies within the mode. The optimized coupling coefficient between samples obtained by recursive optimization The complex dependencies between the internal samples of the mode have been reflected. In order to integrate the internal dependencies into the multimodal coupling analysis, the coupling coefficients between the optimized samples are As a weight coefficient, it is included in the calculation of the coupling matrix between modes.

[0047] S2. Calculate the coupling matrix through the multimodal coupling analysis method, globally optimize the multimodal coupling analysis process, and dynamically adjust the coupling matrix to obtain an updated coupling matrix; obtain the predicted concentration value of the pollutant based on the updated coupling matrix; and detect and analyze the soil of the contaminated site based on the predicted concentration value of the pollutant.

[0048] In order to further utilize the recursive random field model, a multimodal coupling analysis method is proposed; the core of the multimodal coupling analysis method is to construct a coupling matrix to represent the interaction between different modes; by coupling data sets of different modes, the interaction between data sets of different modes is mined; the characteristic vectors of different modes are coupled and analyzed to capture the complex dependencies hidden in the multimodal data sets.

[0049] The multi-mode coupling analysis method is implemented by calculating the coupling matrix. The calculation formula of the coupling matrix is:

[0050] in, is at the time step The resulting coupling matrix represents the interaction between different modes; is modal and The coupling coefficient between is the transformation matrix between modes, used to transform the modes The eigenvectors of all soil samples are mapped to the modal The feature space of Representing modality Datasets; represents transpose; is the bias term; is the hyperbolic cosine function, which is used to capture the nonlinear relationship between modes; is the adjustment weight of the recursive random field. The output of the coupling matrix will be directly used for the next step of optimization of the recursive random field, ensuring that the results of each coupling analysis can be reflected in the subsequent recursive process.

[0051] The entire multimodal coupling analysis process is globally optimized, the recursive optimization objective function is minimized globally, and the overall prediction performance of the recursive random field model is improved. The recursive optimization objective function is as follows:

[0052]

[0053] in, is the recursive optimization objective function, which is minimized in each recursive iteration by adjusting the coupling coefficient between the optimized samples. and the transformation matrix between modes , gradually approaching the optimal solution; is the total number of time steps; and are regularization parameters; is the Frobenius norm. The recursive optimization process aims to ensure that the recursive random field model can accurately model the complex interrelationships in multimodal data sets and improve the prediction accuracy of the recursive random field model through global optimization.

[0054] In order to enhance the adaptability of the recursive random field model, an adaptive coupling mechanism is further introduced to dynamically adjust the coupling matrix so that it can adapt to the changing modal relationship in the recursive process. The adaptive update formula of the coupling matrix is ​​as follows:

[0055]

[0056] in, is at the time step The resulting coupling matrix; is the learning rate, which controls the update amplitude of the coupling matrix; is the recursive optimization objective function Gradients with respect to the coupling matrix; is the adjustment coefficient; It is used to adjust the direction of the coupling matrix. The updated coupling matrix obtained after the coupling matrix is ​​adaptively updated enables the coupling matrix to dynamically adapt to new data patterns during the recursive optimization process, thereby improving flexibility and accuracy.

[0057] The concentration and distribution of pollutants are predicted through regression models. Based on the updated coupling matrix, the data in the multimodal data set are mapped to the pollutant concentration space to achieve accurate prediction of pollutant concentration. The specific regression model formula is as follows:

[0058]

[0059] in, It is The predicted value of each soil sample, that is, the predicted value of the pollutant concentration; are the coefficients in the regression model, indicating the weights of different terms; is the hyperbolic tangent function, capturing nonlinear characteristics; is modal Next The feature vector of each soil sample; is the regularization coefficient. The regression model combines the eigenvectors of soil samples under different modes and the updated coupling matrix to output the predicted concentration of pollutants.

[0060] After obtaining the predicted concentration value of the pollutant, the predicted concentration value of the pollutant is combined with the actual geographical location information of the soil sample obtained from GPS to achieve comprehensive detection and analysis of the soil at the contaminated site. The actual geographical location information of each soil sample is matched with the predicted concentration value of the pollutant, and a pollutant concentration distribution map of the entire contaminated site is generated through an interpolation algorithm (such as Kriging interpolation or inverse distance weighted method), which intuitively displays the spatial distribution of pollutants in the site and helps identify areas with higher pollutant concentrations.

[0061] Based on the generated pollutant concentration distribution map, combined with soil standards and environmental regulations, a risk assessment of pollutant concentrations in different areas is conducted to obtain pollution risk assessment results; a multi-factor decision-making model (such as the AHP hierarchical analysis method or the fuzzy comprehensive evaluation method) is used to assess the potential risks of soil pollution to the environment and health, and classify them into low, medium, and high risk areas; based on the pollution risk assessment results, a soil remediation plan is formulated for high-risk areas.

[0062] In summary, a soil detection and analysis method for contaminated sites was completed.

[0063] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0064] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0065] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for detecting and analyzing soil in contaminated sites. It is characterized in that The following steps are involved: S1. Obtain the data reflecting the physical and chemical properties of soil samples from the contaminated site to obtain a multimodal data set; based on the multimodal data set, construct a recursive random field model. During the construction of the recursive random field model, initialize the recursive random field model to obtain the probability density function of the recursive random field model. The specific implementation formula is as follows: ; in, represents the probability density function of the recursive random field model; represents the target variable; Represents the dataset for each modality; is modal Next Soil samples and Coupling coefficient between soil samples; It is the measurement mode The characteristic vector of the soil sample and A function of the similarity between is modal Next The feature vector of the soil sample With Contaminant concentration labels for soil samples The mapping function between is the number of modal types; The coupling relationship between soil samples is gradually adjusted through recursive optimization. In each recursion, the recursive random field model optimizes the coupling coefficient between samples by minimizing the objective function to obtain the optimized coupling coefficient between samples. The specific implementation formula is as follows: ; in, is the objective function; is the regularization parameter; yes Regularization term; S2. Based on the optimized coupling coefficient between samples, the characteristic vectors of soil samples under different modes are coupled and analyzed by multimodal coupling analysis method to obtain the coupling matrix. The specific calculation formula of the coupling matrix is ​​as follows: ; in, is at the time step The resulting coupling matrix; is modal and The coupling coefficient between is the number of modal types; is the coupling coefficient between the optimized samples; Representing modality Datasets; Representing modality Datasets; is the transformation matrix between modes; represents transpose; is the bias term; are the adjustment weights of the recursive random field model; is the pollutant concentration; is the probability density function of the recursive random field model; The multimodal coupling analysis process is globally optimized. In the process of global optimization of the multimodal coupling analysis process, a recursive optimization objective function is defined and minimized. The specific calculation formula is as follows: ; in, is the recursive optimization objective function; is the total number of time steps; and are regularization parameters; yes norm; An adaptive coupling mechanism is introduced to dynamically adjust the coupling matrix to obtain an updated coupling matrix. The adaptive update formula of the coupling matrix is ​​as follows: ; in, is at the time step The resulting coupling matrix; is the learning rate; is the recursive optimization objective function Gradients with respect to the coupling matrix; is the adjustment coefficient; Used to adjust the direction of the coupling matrix; Based on the updated coupling matrix, the concentration and distribution of pollutants are predicted by the regression model to obtain the predicted concentration value of pollutants. The formula of the specific regression model is as follows: ; in, It is The predicted value of each soil sample, that is, the predicted value of the pollutant concentration; is the coefficient in the regression model; is the hyperbolic tangent function; is modal Next The feature vector of each soil sample; is the regularization coefficient; yes Regularization term; Based on the predicted concentration values ​​of pollutants, the soil of the contaminated site is tested and analyzed.

2. A method for detecting and analyzing soil in a contaminated site according to claim 1, It is characterized in that In the S2, it also includes: Generate a pollutant concentration distribution map based on the predicted pollutant concentration values ​​and the actual geographical location information of the obtained soil samples; conduct a risk assessment of pollutant concentrations in different areas based on the pollutant concentration distribution map to obtain pollution risk assessment results; and formulate a soil remediation plan based on the pollution risk assessment results.