A method and system for rapid dynamic determination of the permeability coefficient of unsaturated soil

Abstract

The application discloses a kind of non-saturated soil permeability coefficient fast dynamic determination method and system, the method includes the following steps: the resistivity data of target area soil, dielectric constant data and temperature data are collected;These data are collected according to uniform time resolution, and observation data are obtained;Based on the data collected, construct soil heterogeneous unit feature vector;The soil heterogeneous unit feature vector is divided, and the initial permeability coefficient field is generated;Rock physics constitutive model is constructed;Fusion observation data and hydrological dynamics model and joint rock physics constitutive model, dynamically update permeability coefficient field, and apply physical constraint to optimize;Finally, output the spatial and temporal distribution of updated permeability coefficient and the uncertainty quantification result;The application can monitor the change of water in non-saturated soil in real time, accurately assess the water distribution of soil and can find the geological disaster risk such as soil seepage, landslide in advance.

Description

Technical Field

[0001] This invention relates to the fields of geotechnical engineering and hydrogeology, specifically to a method and system for rapid dynamic determination of the permeability coefficient of unsaturated soil. Background Technology

[0002] The permeability coefficient of unsaturated soil is a core parameter describing soil moisture transport capacity. Traditional measurement methods, such as instantaneous profiling and centrifugation, suffer from high cost, low efficiency, destructiveness, and insufficient resolution. Existing physical techniques, such as ERT and GPR, can indirectly reflect soil characteristics, but the accuracy of inversion from a single data source is limited, and they lack dynamic update capabilities, failing to achieve real-time measurement with high spatiotemporal resolution. Furthermore, the permeability characteristics of heterogeneous soils are influenced by the coupling of multiple factors, including porosity, temperature, and volumetric water content. Existing static models do not consider the dynamic coupling effects of these multiple factors and lack quantitative assessment of the reliability of the inversion results. Therefore, it is necessary to integrate multi-source geophysical data with hydrogeological measurement data to solve the challenge of high-precision spatiotemporal dynamic measurement of the permeability coefficient of unsaturated soil. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a rapid dynamic measurement method and system for the permeability coefficient of unsaturated soil. The purpose is to achieve rapid and high-precision dynamic measurement of the permeability coefficient of unsaturated soil by using multi-source physical data fusion, fuzzy C-means clustering, and ensemble Kalman filtering dynamic assimilation technology. This overcomes the limitations of traditional measurement methods, improves the measurement speed and accuracy of permeability coefficient, and enables dynamic monitoring of soil moisture characteristics.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for rapid dynamic determination of the permeability coefficient of unsaturated soil, comprising the following steps:

[0005] Step S1: Obtain resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data, and temperature data of the soil in the target area are collected at a uniform time resolution to obtain observation data;

[0006] Step S2: Construct feature vectors for soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data;

[0007] Step S3: Based on the fuzzy C-means clustering algorithm, the feature vectors of soil heterogeneous units are divided to identify and classify different soil properties, and at the same time, the initial permeability coefficient field of the soil in the target area is generated.

[0008] Step S4: Construct a rock physical constitutive model and establish the nonlinear relationship between the permeability coefficient and soil physical parameters in the target area, including the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil.

[0009] Step S5: Based on the permeability coefficient, soil temperature and soil physical parameters, an initial set is generated by an adaptive ensemble Kalman filter algorithm. The initial set is input into the hydrodynamic model to generate a prediction set. Combined with the rock physical constitutive model, the prediction set is mapped to the observation space to obtain the soil resistivity data, dielectric constant data and predicted values ​​of permeability coefficient of the soil in the target area at different times. The prediction set is dynamically updated based on the predicted values ​​and observation data.

[0010] Step S6: Based on the updated prediction set, output the spatiotemporal distribution of the permeability coefficient in the prediction set and the uncertainty quantification results.

[0011] Furthermore, the specific process of step S3 is as follows: Establish the objective function of the fuzzy C-means clustering algorithm. , represented as:

[0012] (1);

[0013] In the formula, This represents the total number of eigenvectors of heterogeneous soil units; Indicates the number of clusters; Indicates the first The feature vector of the soil heterogeneous unit belongs to the th soil heterogeneous unit. Membership degree of a class; Indicates the fuzzy index; Indicates the first Feature vectors of soil heterogeneous units; Indicates the first Cluster centers of a class; Indicates the space regularization weights; Indicates the first The neighborhood of the feature vector of a soil heterogeneous unit; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers Indicates the first The index of the feature vector of other soil heterogeneous units in the neighborhood of the feature vector of a soil heterogeneous unit;

[0014] By alternately optimizing membership degree and cluster center Once convergence is achieved, the clustering results are mapped to the initial permeability field of the soil in the target region. , represented as:

[0015] (2);

[0016] In the formula, Indicates the soil of the target area The initial permeability field at each location; Indicates the first Typical values ​​of the penetration coefficient for clustering.

[0017] Furthermore, the specific process of step S4 is as follows: Resistivity data of the soil in the target area are established respectively. Dielectric constant data and the initial permeability field medium permeability Soil physical parameters, including porosity saturation and volumetric water content The quantitative relationship is expressed as:

[0018] (3);

[0019] (4);

[0020] (5);

[0021] In the formula, Indicates the cementing factor; , These represent the porosity index and the saturation index, respectively. Indicates volumetric water content; This represents a constant related to particle shape; Indicates the saturation index; Indicates the density of water;

[0022] Constructing a multi-parameter joint inversion equation: Combine equations (3), (4), and (5) to form a closed equation set, and input the resistivity data of the soil in the target area. Dielectric constant data Porosity was solved iteratively using the Newton-Raphson method. and saturation Then the solution and Substitute into equation (5) to determine the permeability coefficient. The computational model.

[0023] Furthermore, the specific process of step S5 is as follows:

[0024] Based on the soil porosity of the target area Volumetric water content Permeability coefficient and soil temperature Construct a system that includes soil dynamic state variables ( , , ) and static parameters ( The initial state vector of ) :

[0025] (6);

[0026] In the formula, Indicates transpose;

[0027] Based on the initial state vector Generate an initial set, including A set member, a set member In the initial state vector The result is obtained by adding random perturbations.

[0028] The hydrodynamic model is used to predict the state of all ensemble members. Specifically, the hydrological model operators are used to predict the state of each ensemble member at the next time step, forming a prediction set, represented as follows:

[0029] (7);

[0030] In the formula, Indicates the first Each set member in The state at any given moment; Indicates the first Each set member in The state at any given moment; For hydrological model operators; For time step;

[0031] By combining a rock physics constitutive model and mapping the prediction set to the observation space, soil resistivity data of the target area at different times are obtained. Dielectric constant data Permeability coefficient Predicted value , represented as:

[0032] (8);

[0033] In the formula, For rock physics model; This represents the set prediction generated by the hydrological model operator;

[0034] Comparison with predicted values With observation data And calculate the Kalman gain matrix. The prediction set is updated using the Kalman gain matrix, as follows:

[0035] (9);

[0036] (10);

[0037] In the formula, The prediction error covariance matrix; This is the observation error; , These are the prediction sets before and after the update, respectively; express transpose;

[0038] The updated prediction set is fed back to the hydrological model operator for simulation at the next time step;

[0039] Simultaneously, covariance inflation and localization are introduced during the inversion process of the adaptive ensemble Kalman filter algorithm, and physical constraints are applied; the process of the adaptive ensemble Kalman filter algorithm is repeated until the distribution of predicted values ​​matches that of observed values; the convergence criterion is set to the predicted value. With observation data The distributional Nash efficiency coefficient NSE > 0.8, and the change over three consecutive iterations is < 1%.

[0040] Furthermore, the specific process of step S6 is as follows: using the Kriging interpolation method, the discrete permeability coefficients in the updated prediction set are mapped to a continuous spatiotemporal grid; and through statistical indicators and probability distributions, the uncertainty of the permeability coefficients is quantified, and the probability of the permeability coefficients exceeding the threshold is calculated to indicate potential seepage risks.

[0041] Furthermore, the uncertainty entropy of the permeability coefficient is calculated using equation (11), and the probability that the permeability coefficient exceeds the threshold is calculated using equation (12), expressed as:

[0042] (11);

[0043] (12);

[0044] In the formula, Permeability coefficient exist Uncertain entropy at any given moment; Indicates in Permeability coefficient at any time Category The probability of; express The value is greater than the permeability coefficient threshold, i.e., the preset critical value. The probability of; Indicates the first The penetration coefficient of each sample; The total number of categories representing the penetration coefficient; the samples are set members in the updated prediction set.

[0045] Furthermore, before constructing the feature vector of the soil heterogeneous unit based on the collected resistivity data, dielectric constant data, and temperature data, the resistivity data, dielectric constant data, and temperature data are preprocessed, including denoising and normalization.

[0046] A rapid dynamic measurement system for the permeability coefficient of unsaturated soil includes:

[0047] The data acquisition module is used to acquire resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data, and temperature data of the soil in the target area are acquired at a uniform time resolution to obtain observation data;

[0048] The feature vector construction module is used to construct feature vectors for soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data.

[0049] The partitioning module is used to partition the feature vectors of heterogeneous soil units based on the fuzzy C-means clustering algorithm in order to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area.

[0050] The model building module is used to build a rock physics constitutive model and establish the nonlinear relationship between the permeability coefficient and soil physical parameters in the target area, including the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil.

[0051] The update and optimization module is used to generate an initial set based on permeability coefficient, soil temperature and soil physical parameters through an adaptive ensemble Kalman filter algorithm. The initial set is then input into a hydrodynamic model to generate a prediction set. Combined with a rock physics constitutive model, the prediction set is mapped to the observation space to obtain soil resistivity data, dielectric constant data and predicted values ​​of permeability coefficient at different times in the target area. The prediction set is then dynamically updated based on the predicted values ​​and observation data.

[0052] The output module is used to output the spatiotemporal distribution of the permeability coefficients in the updated prediction set and the uncertainty quantification results.

[0053] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a method for rapid dynamic determination of the permeability coefficient of unsaturated soil.

[0054] A non-volatile computer storage medium storing computer-executable instructions that execute a method for rapid dynamic determination of the permeability coefficient of unsaturated soil.

[0055] Compared with existing technologies, the present invention has the following advantages:

[0056] (1) This invention achieves high-precision, high-efficiency and high-resolution determination of the permeability coefficient of unsaturated soil through the synergistic innovation of multi-source data fusion, FCM clustering, dynamic assimilation algorithm and set statistics. At the same time, it provides uncertainty quantification and risk warning functions, significantly improving the dynamic characterization ability of the seepage process of complex heterogeneous soil, and providing scientific theoretical support for geological disaster prevention and agricultural water conservation.

[0057] (2) This invention supports hourly updates of unsaturated permeability coefficients through a dynamic assimilation algorithm, quickly capturing the dynamic changes of transient hydrological events, overcoming the shortcomings of traditional methods that can only estimate permeability coefficients statically and in isolation, and significantly improving the timeliness of monitoring and prediction.

[0058] (3) The present invention achieves spatial resolution down to the centimeter level through a multi-source data fusion algorithm, which can finely characterize soil heterogeneity and solve the problem of local detail loss caused by traditional homogeneity assumptions and isolated inversion.

[0059] (4) This invention quantifies uncertainty through set statistics, provides a confidence interval for risk decision-making, and makes up for the limitation of a single solution lacking credibility assessment. Attached Figure Description

[0060] Figure 1 This is a flowchart of the method of the present invention.

[0061] Figure 2 This is a schematic diagram of data acquisition according to the present invention.

[0062] Figure 3 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0063] like Figure 1 As shown, the present invention provides a technical solution: a method for rapid dynamic determination of the permeability coefficient of unsaturated soil, comprising the following steps:

[0064] Step S1: Obtain soil resistivity data for the target area. Dielectric constant data and temperature data The soil in the target area is unsaturated; resistivity, dielectric constant, and temperature data of the soil in the target area are collected at a uniform time resolution to obtain observation data. .

[0065] Specifically, high-density resistivity (ERT), ground-penetrating radar (GPR), and soil temperature sensors were used to acquire soil resistivity data in the target area. Dielectric constant data and temperature data ;like Figure 2 As shown, Figure 2 In the diagram, 1 represents the soil in the target area (unsaturated soil), 2 represents the high-density electrical resistivity tomography (EDT) line, 3 represents the ground-penetrating radar antenna, 4 represents the soil temperature sensor, 5 represents the main unit of the high-density EDT instrument, 6 represents the soil temperature sensor receiver, 7 represents the ground-penetrating radar main unit, and 8 represents the data processing PC. Resistivity data of the soil (unsaturated soil) in the target area are collected using the high-density EDT line 2, the ground-penetrating radar antenna 3, and the soil temperature sensor 4, respectively. Dielectric constant data and temperature data The data is received by the high-density electrical resistivity meter host 5, the soil temperature sensor receiver 6, and the ground-penetrating radar host 7, and finally transmitted to the data processing PC to complete the collection.

[0066] Step S2: Based on the collected resistivity data Dielectric constant data and temperature data Constructing feature vectors of soil heterogeneous units , This indicates transpose.

[0067] Among them, based on the collected resistivity data Dielectric constant data and temperature data Constructing feature vectors of soil heterogeneous units Previously, resistivity data could be analyzed. Dielectric constant data and temperature data Preprocessing is performed (to eliminate the dimensional differences between them). Preprocessing includes denoising and normalization: , , ; Resistivity data after preprocessing ; Indicates the first One resistivity data point; This represents the mean of the resistivity data; This represents the standard deviation of resistivity data; This represents the preprocessed dielectric constant data. ; Indicates the first Dielectric constant data ; This represents the mean of the electrical constant data; The standard deviation of dielectric constant data; This indicates the temperature data after preprocessing. ; Indicates the first Temperature data ; This represents the mean of the temperature data; This represents the standard deviation of the temperature data.

[0068] Step S3: Based on the fuzzy C-means (FCM) clustering algorithm, analyze the feature vectors of soil heterogeneous units. The soil is divided into sections to identify and classify different soil properties, while simultaneously generating an initial permeability field for the target area. .

[0069] The specific process of step S3 is as follows: Establish the objective function of the fuzzy C-means (FCM) clustering algorithm. , represented as:

[0070] (1);

[0071] In the formula, This represents the total number of eigenvectors of heterogeneous soil units; Indicates the number of clusters; Indicates the first The feature vector of the soil heterogeneous unit belongs to the th soil heterogeneous unit. Membership degree of a class; This represents the fuzziness index, which is usually taken as 2. Indicates the first Feature vectors of soil heterogeneous units; Indicates the first Cluster centers of a class; Indicates the space regularization weights; Indicates the first The neighborhood of the feature vector of a soil heterogeneous unit; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers Indicates the first The index of the eigenvectors of other soil heterogeneous units in the neighborhood of the eigenvector of a given soil heterogeneous unit.

[0072] The objective function of the fuzzy C-means clustering algorithm designed in this invention innovatively introduces a spatial weight matrix, namely the second term in equation (1), to force adjacent units to have similar membership degrees and suppress "chessboard" artifacts.

[0073] By alternately optimizing membership degree and cluster center Once convergence is achieved, the clustering results are mapped to the initial permeability field of the soil in the target region. , represented as:

[0074] (2);

[0075] In the formula, Indicates the soil of the target area The initial permeability field at each location; Indicates the first Typical values ​​of the permeability coefficient for clustering can be determined based on prior knowledge, experimental data, or experience.

[0076] Step S4: Construct a rock physical constitutive model and establish soil resistivity data for the target area. Dielectric constant data and the initial permeability field Permeability coefficient and soil physical parameters (porosity) saturation and volumetric water content The nonlinear relationship of ).

[0077] The specific process of step S4: First, based on Archie's law (Equation (3)), Topp's formula (Equation (4)) and Kozeny-Carman's equation (Equation (5)), the resistivity data of the soil in the target area are established respectively. Dielectric constant data and the initial permeability field medium permeability With soil physical parameters (porosity) saturation and volumetric water content The quantitative relationship between them is expressed as:

[0078] (3);

[0079] (4);

[0080] (5);

[0081] In the formula, Indicates cementing factor (sand) clay ); , These represent the porosity index and saturation index, respectively (usually...). , ); Indicates volumetric water content; This represents a constant related to particle shape; Indicates the saturation index; This indicates the density of water.

[0082] Constructing a multi-parameter joint inversion equation: Combine equations (3), (4), and (5) to form a closed equation set, and input the resistivity data of the soil in the target area. Dielectric constant data Porosity was solved iteratively using the Newton-Raphson method. and saturation Then the solution and Substitute into equation (5) to determine the permeability coefficient. The calculation model for permeability coefficient can quickly and easily calculate the soil permeability coefficient.

[0083] Step S5: Based on permeability coefficient Soil temperature With soil physical parameters (porosity) saturation and volumetric water content An initial set is generated using the Adaptive Ensemble Kalman Filter (AEnKF) algorithm. This initial set is then input into a hydrodynamic model to generate a prediction set. Combined with a rock physics constitutive model (Equations (3) to (5)), the prediction set is mapped to the observation space to obtain soil resistivity data of the target area at different times. Dielectric constant data Permeability coefficient Predicted value Based on predicted values and observation data The prediction set is updated dynamically.

[0084] The specific process of step S5 is as follows:

[0085] Soil porosity obtained from the above steps Volumetric water content Permeability coefficient and soil temperature Construct a system that includes soil dynamic state variables ( , , ) and static parameters ( The initial state vector of ) :

[0086] (6);

[0087] In the formula, This indicates transpose.

[0088] Based on the initial state vector Generate an initial set to characterize uncertainty, including A set of members (usually) =50~200), set members In the initial state vector The result is obtained by adding random perturbations.

[0089] The hydrodynamic model is run to predict the state of all ensemble members. Specifically, hydrological model operators (such as Hydrus-3D software) are used to predict the state of each ensemble member at the next time step, forming a prediction set, represented as:

[0090] (7);

[0091] In the formula, Indicates the first Each set member in The state at any given moment; Indicates the first Each set member in The state at any given moment; For hydrological model operators (such as those in Hydrus-3D software); The time step is typically the temporal resolution of the observed data.

[0092] By combining the rock physics constitutive model (Equations (3) to (5)), the prediction set is mapped to the observation space to obtain soil resistivity data of the target area at different times. Dielectric constant data Permeability coefficient Predicted value , represented as:

[0093] (8);

[0094] In the formula, For rock physics constitutive models, referring to Archie's law (Equation (3)), Topp's formula (Equation (4)) and Kozeny-Carman's equation (Equation (5)); This represents the set prediction generated by the hydrological model operator.

[0095] Comparison with predicted values With observation data And calculate the Kalman gain matrix. The prediction set is updated using the Kalman gain matrix, as follows:

[0096] (9);

[0097] (10);

[0098] In the formula, The prediction error covariance matrix; This is the observation error; , These are the prediction sets before and after the update, respectively; express The transpose of .

[0099] The updated prediction set is fed back to the hydrological model operator (such as Hydrus-3D software) for simulation at the next time step.

[0100] Meanwhile, covariance inflation (inflation factor = 1.05~1.3) and localization (using the Gaspari-Cohn function) are introduced during the inversion process of the Adaptive Ensemble Kalman Filter (AEnKF) algorithm to alleviate ensemble sampling errors. Physical constraints are also applied (the upper and lower limits of the permeability coefficient, which in this embodiment can be set to (10...). −10 ~10 -3 (m / s), avoiding physically unreasonable values ​​(such as negative permeability coefficients) and suppressing artifacts. Repeat the above process until the predicted value is reached. With observation data The distribution is consistent, and the convergence criterion is set as a global indicator, i.e., the predicted value. With observation data The inter-distribution Nash efficiency coefficient (NSE) is greater than 0.8, and the change over three consecutive iterations is less than 1%; the predicted value... With observation data The inter-distribution Nash efficiency coefficient NSE can be expressed as: ; Indicates the first Of the predicted values; Indicates the first One observation data; This represents the average value of the observed data; This indicates the quantity of observed data or predicted values.

[0101] Step S6: Based on the updated prediction set, output the spatiotemporal distribution of the permeability coefficient in the prediction set and the uncertainty quantification results.

[0102] The specific process of step S6 is as follows: using the Kriging interpolation method, the discrete permeability coefficients in the updated prediction set are mapped to a continuous spatiotemporal grid; and through statistical indicators and probability distributions, the uncertainty of the permeability coefficients is quantified, and the probability of the permeability coefficients exceeding the threshold is calculated to indicate potential seepage risks.

[0103] The uncertainty entropy of the permeability coefficient is calculated using equation (11), and the probability that the permeability coefficient exceeds the threshold is calculated using equation (12), which is expressed as:

[0104] (11);

[0105] (12);

[0106] In the formula, Permeability coefficient exist Uncertain entropy at any given moment; Indicates in Permeability coefficient at any time Category The probability of; express Greater than the permeability coefficient threshold (preset critical value) The probability of; Indicates the first The penetration coefficient of each sample (a member of the updated prediction set); It represents the total number of all possible states or categories of the permeability coefficient.

[0107] like Figure 3 As shown, a rapid dynamic measurement system for the permeability coefficient of unsaturated soil is characterized by comprising:

[0108] The data acquisition module is used to acquire resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data, and temperature data of the soil in the target area are acquired at a uniform time resolution to obtain observation data.

[0109] The feature vector construction module is used to construct feature vectors for soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data.

[0110] The partitioning module is used to partition the feature vectors of heterogeneous soil units based on the fuzzy C-means clustering algorithm in order to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area.

[0111] The model building module is used to construct a rock physics constitutive model and establish the nonlinear relationship between the permeability coefficient and soil physical parameters in the target area, including the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil.

[0112] The update and optimization module is used to generate an initial set based on permeability coefficient, soil temperature and soil physical parameters through an adaptive ensemble Kalman filter algorithm. The initial set is then input into a hydrodynamic model to generate a prediction set. Combined with a rock physics constitutive model, the prediction set is mapped to the observation space to obtain soil resistivity data, dielectric constant data and predicted values ​​of permeability coefficient at different times in the target area. The prediction set is then dynamically updated based on the predicted values ​​and observation data.

[0113] The output module is used to output the spatiotemporal distribution of the permeability coefficients in the updated prediction set and the uncertainty quantification results.

[0114] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a method for rapid dynamic determination of the permeability coefficient of unsaturated soil.

[0115] A non-volatile computer storage medium storing computer-executable instructions that execute a method for rapid dynamic determination of the permeability coefficient of unsaturated soil.

[0116] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A rapid dynamic method for determining the permeability coefficient of unsaturated soil, characterized in that, Includes the following steps: Step S1: Obtain resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data, and temperature data of the soil in the target area are collected at a uniform time resolution to obtain observation data; Step S2: Construct feature vectors for soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data; Step S3: Based on the fuzzy C-means clustering algorithm, the feature vectors of soil heterogeneous units are divided to identify and classify different soil properties, and at the same time, the initial permeability coefficient field of the soil in the target area is generated. Step S4: Construct a rock physical constitutive model and establish the nonlinear relationship between the permeability coefficient and soil physical parameters in the target area, including the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil. Step S5: Based on the permeability coefficient, soil temperature and soil physical parameters, an initial set is generated by an adaptive ensemble Kalman filter algorithm. The initial set is input into the hydrodynamic model to generate a prediction set. Combined with the rock physical constitutive model, the prediction set is mapped to the observation space to obtain the soil resistivity data, dielectric constant data and predicted values ​​of permeability coefficient of the soil in the target area at different times. The prediction set is dynamically updated based on the predicted values ​​and observation data. Step S6: Based on the updated prediction set, output the spatiotemporal distribution of the penetration coefficient and the uncertainty quantification results in the prediction set; The specific process of step S4: Establish the resistivity data of the soil in the target area. Dielectric constant data and the initial permeability field medium permeability Soil physical parameters, including porosity saturation and volumetric water content The quantitative relationship is expressed as: （3）； （4）； （5）； In the formula, Indicates the cementing factor; , These represent the porosity index and the saturation index, respectively. Indicates volumetric water content; This represents a constant related to particle shape; Indicates the saturation index; Indicates the density of water; Constructing a multi-parameter joint inversion equation: Combine equations (3), (4), and (5) to form a closed equation set, and input the resistivity data of the soil in the target area. Dielectric constant data Porosity was solved iteratively using the Newton-Raphson method. and saturation Then the solution and Substitute into equation (5) to determine the permeability coefficient. The computational model; The specific process of step S5 is as follows: Based on the soil porosity of the target area Volumetric water content Permeability coefficient and soil temperature Construct a system that includes soil dynamic state variables ( , , ) and static parameters ( The initial state vector of ) : （6）； In the formula, Indicates transpose; Based on the initial state vector Generate an initial set, including A set member, a set member In the initial state vector The result is obtained by adding random perturbations. The hydrodynamic model is used to predict the state of all ensemble members. Specifically, the hydrological model operators are used to predict the state of each ensemble member at the next time step, forming a prediction set, represented as follows: （7）； In the formula, Indicates the first Each set member in The state at any given moment; Indicates the first Each set member in The state at any given moment; For hydrological model operators; For time step; By combining a rock physics constitutive model and mapping the prediction set to the observation space, soil resistivity data of the target area at different times are obtained. Dielectric constant data Permeability coefficient Predicted value , represented as: （8）； In the formula, For rock physics constitutive models; This represents the set prediction generated by the hydrological model operator; Comparison with predicted values With observation data And calculate the Kalman gain matrix. The prediction set is updated using the Kalman gain matrix, as follows: （9）； （10）； In the formula, The prediction error covariance matrix; This is the observation error; , These are the prediction sets before and after the update, respectively; express transpose; The updated prediction set is fed back to the hydrological model operator for simulation at the next time step; Simultaneously, covariance inflation and localization are introduced during the inversion process of the adaptive ensemble Kalman filter algorithm, and physical constraints are applied; the process of the adaptive ensemble Kalman filter algorithm is repeated until the distribution of predicted values ​​matches that of observed values; the convergence criterion is set to the predicted value. With observation data The distributional Nash efficiency coefficient NSE > 0.8, and the change over three consecutive iterations is < 1%.

2. The method for rapid dynamic determination of the permeability coefficient of unsaturated soil according to claim 1, characterized in that: The specific process of step S3 is as follows: Establish the objective function of the fuzzy C-means clustering algorithm. , represented as: （1）； In the formula, This represents the total number of eigenvectors of heterogeneous soil units; Indicates the number of clusters; Indicates the first The feature vector of the soil heterogeneous unit belongs to the th soil heterogeneous unit. Membership degree of a class; Indicates the fuzzy index; Indicates the first Feature vectors of soil heterogeneous units; Indicates the first Cluster centers of a class; Indicates the space regularization weights; Indicates the first The neighborhood of the feature vector of a soil heterogeneous unit; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers; Indicates the first The eigenvector of the soil heterogeneous unit and the eigenvector of the soil heterogeneous unit Membership degree between cluster centers Indicates the first The index of the feature vector of other soil heterogeneous units in the neighborhood of the feature vector of a soil heterogeneous unit; By alternately optimizing membership degree and cluster center Once convergence is achieved, the clustering results are mapped to the initial permeability field of the soil in the target region. , represented as: （2）； In the formula, Indicates the soil of the target area The initial permeability field at each location; Indicates the first Typical values ​​of the penetration coefficient for clustering.

3. The method for rapid dynamic determination of the permeability coefficient of unsaturated soil according to claim 2, characterized in that: The specific process of step S6 is as follows: using the Kriging interpolation method, the discrete permeability coefficients in the updated prediction set are mapped to a continuous spatiotemporal grid; and through statistical indicators and probability distributions, the uncertainty of the permeability coefficients is quantified, and the probability of the permeability coefficients exceeding the threshold is calculated to indicate potential seepage risks.

4. The method for rapid dynamic determination of the permeability coefficient of unsaturated soil according to claim 3, characterized in that: The uncertainty entropy of the permeability coefficient is calculated using equation (11), and the probability of the permeability coefficient exceeding the threshold is calculated using equation (12), expressed as: （11）； （12）； In the formula, Permeability coefficient exist Uncertain entropy at any given moment; Indicates in Permeability coefficient at any time Category The probability of; express The value is greater than the permeability coefficient threshold, i.e., the preset critical value. The probability of; Indicates the first The penetration coefficient of each sample; The total number of categories representing the penetration coefficient; the samples are set members in the updated prediction set.

5. The method for rapid dynamic determination of the permeability coefficient of unsaturated soil according to claim 4, characterized in that: Before constructing the feature vector of soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data, the resistivity data, dielectric constant data, and temperature data are preprocessed, including denoising and normalization.

6. A rapid dynamic measurement system for the permeability coefficient of unsaturated soil, applied to the rapid dynamic measurement method for the permeability coefficient of unsaturated soil as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to acquire resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil. The resistivity, dielectric constant, and temperature data of the soil in the target area were collected at a uniform time resolution to obtain the observation data. The feature vector construction module is used to construct feature vectors for soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data. The partitioning module is used to partition the feature vectors of heterogeneous soil units based on the fuzzy C-means clustering algorithm in order to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area. The model building module is used to build a rock physics constitutive model and establish the nonlinear relationship between the permeability coefficient and soil physical parameters in the target area, including the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil. The update and optimization module is used to generate an initial set based on permeability coefficient, soil temperature and soil physical parameters through an adaptive ensemble Kalman filter algorithm. The initial set is then input into a hydrodynamic model to generate a prediction set. Combined with a rock physics constitutive model, the prediction set is mapped to the observation space to obtain soil resistivity data, dielectric constant data and predicted values ​​of permeability coefficient at different times in the target area. The prediction set is then dynamically updated based on the predicted values ​​and observation data. The output module is used to output the spatiotemporal distribution of the permeability coefficients in the updated prediction set and the uncertainty quantification results.

7. An electronic device, characterized in that, The device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a rapid dynamic determination method for the permeability coefficient of unsaturated soil as described in any one of claims 1-5.

8. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer can execute instructions to perform a rapid dynamic determination method for the permeability coefficient of unsaturated soil as described in any one of claims 1-5.

Images