Soil structure optimization method based on CT scanning and hydrothermal transmission characteristic analysis

Abstract

The invention discloses a soil structure optimization method based on CT scanning and hydrothermal transmission characteristic analysis, and relates to the technical field of soil structure optimization. The method comprises the following steps: scanning a soil sample by utilizing industrial CT equipment, extracting pore structure parameters of a scanned image, researching hydrothermal transmission characteristics of the soil sample by utilizing numerical simulation and experiments, obtaining hydrothermal transmission characteristic parameters, constructing a LightGBM model by taking the pore structure parameters as input characteristics and taking the hydrothermal transmission characteristic parameters as prediction parameters, and constructing the LightGBM model. Acquiring importance scores of input characteristics through the trained LightGBM model, identifying pore structure parameter types having significant influence on hydrothermal transmission parameters as optimization objects, and constructing a soil structure optimization model used in a target area by taking the optimal hydrothermal transmission characteristic of a soil sample as an optimization target; and obtaining an optimal value of the optimization object, and further forming a soil structure optimization strategy.

Description

Technical Field

[0001] This invention relates to the field of soil structure optimization technology, and in particular to a soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis. Background Technology

[0002] Soil is a vital component of the ecosystem, and its structural characteristics directly influence plant growth, water retention, nutrient cycling, and ecosystem stability. Good soil structure improves soil aeration, water retention, and fertility, promotes root growth and development, and reduces soil erosion and nutrient loss. However, due to natural factors (such as weathering and erosion) and human activities (such as over-cultivation and improper irrigation), soil structure is often damaged, leading to a decline in soil quality and consequently impacting agricultural production and the sustainable development of the ecological environment. Therefore, optimizing soil structure is of significant practical importance.

[0003] Traditional methods for optimizing soil structure mainly rely on empirical measures, such as deep tillage, application of organic fertilizers, and planting green manure. While these methods can improve soil structure to some extent, they cannot accurately understand the soil's internal pore structure and water and heat transfer characteristics. Optimization measures are often based on macroscopic experience, making it difficult to precisely control the soil's microstructure. Furthermore, traditional methods typically focus only on a single soil characteristic (such as porosity or fertility) while neglecting the complex interaction between soil structure and water and heat transfer characteristics, thus failing to achieve comprehensive optimization of soil structure.

[0004] In recent years, with the rapid development of computed tomography (CT) technology, its application in soil structure research has become increasingly widespread. CT scans can provide high-resolution images of the soil's internal structure, and through image processing and analysis, detailed information about the soil's pore structure can be accurately obtained. Furthermore, water and heat transfer characteristic analysis techniques, through numerical simulation and experimental studies, can provide in-depth analysis of the transfer processes of water and heat in the soil and their relationship with pore structure. These technological advancements have provided new ideas and methods for soil structure optimization.

[0005] Although CT scanning and hydrothermal transfer characteristic analysis have yielded certain results in soil research, there is currently no systematic method that organically combines these two techniques for soil structure optimization. Existing research mostly focuses on the application of single techniques, lacking a comprehensive analysis and optimization strategy for the complex relationship between soil structure and hydrothermal transfer characteristics. Therefore, this paper proposes a soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis. Summary of the Invention

[0006] The main objective of this invention is to provide a soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis. By accurately acquiring soil pore structure information, deeply analyzing the relationship between pore structure and hydrothermal transfer characteristics, and utilizing multi-objective optimization technology, the method achieves precise optimization of soil structure, improves soil hydrothermal utilization efficiency and ecological function, provides scientific basis and technical support for agricultural production and ecological environmental protection, and can effectively solve the problems in the background technology.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] Soil structure optimization methods based on CT scanning and hydrothermal transfer characteristic analysis include:

[0009] Soil samples from the target area were obtained and processed. The processed soil samples were scanned using an industrial CT scanner. The images obtained from the CT scans were preprocessed. The CT images were segmented using image processing software. The pore and solid phases were distinguished based on the difference in gray values, and the pore structure parameters were extracted. Combined with the pore structure parameters obtained from the CT scans, the hydrothermal transfer characteristics of the soil samples were studied through numerical simulation and experiments to obtain the hydrothermal transfer characteristic parameters of the soil samples.

[0010] Using the pore structure parameters as input features and the water and heat transfer characteristic parameters as prediction parameters, a LightGBM model is constructed. The model is initialized using the LightGBM library, and the basic parameters of the LightGBM model are set. The hyperparameters of the model are tuned using grid search or random search methods. The LightGBM model is trained until its performance index reaches the expected value. The importance score of each input feature is obtained through the trained LightGBM model to identify the types of pore structure parameters that have a significant impact on the water and heat transfer parameters.

[0011] Using the pore structure parameter type with significant influence as the optimization object, the optimal hydrothermal transfer characteristics of the soil sample as the optimization objective, and the value range of the optimization object as the constraint, a soil structure optimization model for the target area is constructed. The optimal value of the optimization object is obtained using the soil structure optimization model, thereby forming a soil structure optimization strategy.

[0012] Furthermore, the pore structure parameters include one or more of the following: porosity, pore size distribution, pore connectivity, pore tortuosity, and pore volume fraction.

[0013] Furthermore, the hydrothermal transport characteristic parameters include moisture transport characteristic parameters, heat transport characteristic parameters, and hydrothermal coupling characteristic parameters, wherein:

[0014] The water transport characteristic parameters include one or more of the following: saturated hydraulic conductivity, unsaturated hydraulic conductivity, soil water potential, and wilting coefficient.

[0015] The heat transfer characteristic parameters include one or more of the following: thermal conductivity, thermal diffusivity, soil temperature gradient, and heat capacity.

[0016] The hydrothermal coupling characteristic parameters include phase change heat.

[0017] Furthermore, the basic parameters of the LightGBM model include the learning rate, the depth of the tree, and the number of leaf nodes; the performance metrics of the LightGBM model include either the mean squared error or the coefficient of determination.

[0018] Furthermore, the identification process for the pore structure parameter type that has a significant impact on the hydrothermal transfer parameters includes the following steps:

[0019] The importance score of the input feature to the prediction parameter is obtained through the trained LightGBM model, denoted as . It is represented as the importance score of the input feature described in the i-th item to the prediction parameter described in the j-th item;

[0020] Define the parameter filtering criteria, including:

[0021] Filtering criterion 1: The importance score of any input feature described in the i-th item to the prediction parameter described in the j-th item. Not less than the first screening threshold;

[0022] Filtering condition 2: For any j-th prediction parameter, the sum of the importance scores of the selected input features is not less than the second filtering threshold;

[0023] Filtering condition 3: For the input feature described in item i, the sum of its importance scores to all the predicted parameters is not less than the third filtering threshold;

[0024] Based on the screening conditions described above, a screening model is constructed for the pore structure parameter types that have a significant impact on the hydrothermal transfer parameters. The screening model is then used to select the input features that meet the criteria as parameter types that have a significant impact on the hydrothermal transfer parameters.

[0025] Furthermore, the expression for the screening model is:

[0026]

[0027] Where n is the number of types of prediction parameters; m is the number of types of selected input features; k1 is the first screening threshold; k2 is the second screening threshold; and k3 is the third screening threshold.

[0028] Furthermore, the range of the first screening threshold is k1∈(0,0.1];

[0029] The second screening threshold ranges from k2 to [0.9, 1].

[0030] The value range of the third screening threshold is k3∈[0.1,0.3].

[0031] The present invention has the following beneficial effects:

[0032] Compared with existing technologies, this solution acquires soil samples from the target area, scans the processed soil samples using industrial CT equipment, extracts pore structure parameters from the scanned images, and studies the hydrothermal transfer characteristics of the soil samples through numerical simulation and experiments. Using these pore structure parameters as input features and the hydrothermal transfer characteristics as prediction parameters, a LightGBM model is constructed. After training, the importance score of each input feature is obtained from the LightGBM model, identifying the types of pore structure parameters that significantly influence hydrothermal transfer. These significantly influential pore structure parameter types are then selected as optimization objects, with the optimal hydrothermal transfer characteristics of the soil sample as the optimization objective. The range of values ​​for the optimization objects is used as constraints to construct a soil structure optimization model for the target area. The optimal values ​​for the optimization objects are obtained using this model, thus forming a soil structure optimization strategy. By combining CT scanning and hydrothermal transfer characteristic analysis techniques, precise soil pore structure information is obtained, the relationship between pore structure and hydrothermal transfer characteristics is analyzed in depth, and multi-objective optimization techniques are used to achieve precise optimization of soil structure, improving soil hydrothermal utilization efficiency and ecological functions, providing scientific basis and technical support for agricultural production and ecological environmental protection. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating the soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to the present invention. Detailed Implementation

[0034] The present invention will be further described below with reference to specific embodiments. The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the present invention. In order to better illustrate the specific embodiments of the present invention, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size.

[0035] The specific implementation process of the technical solution of this invention includes the following steps:

[0036] Step 1: Obtain soil samples from the target area.

[0037] The specific process can be divided into the following steps:

[0038] 1) Soil sample collection and preparation

[0039] Sampling Location and Method: Select representative soil sampling locations. Collect soil samples at different depths using appropriate sampling tools (such as soil samplers).

[0040] 2) Sample processing: The collected soil samples are processed, such as air drying, grinding, and sieving, to remove impurities and ensure the homogeneity of the samples.

[0041] Step 2: Obtain the pore structure parameters of the soil sample.

[0042] The prepared soil samples were scanned using industrial CT equipment. The images obtained from the CT scans were preprocessed, and the CT images were segmented using image processing software. The pore and solid phases were distinguished based on the differences in gray values, and the pore structure parameters were extracted.

[0043] The specific process is as follows:

[0044] 1) Sample scanning was performed using industrial CT equipment. Scanning parameters such as scanning current, voltage, and resolution were set.

[0045] 2) Before scanning, an aluminum filter of a certain thickness is placed between the X-ray source and the sample to prevent artifacts from appearing in the scanning results.

[0046] 3) During scanning, the sample is placed vertically on the turntable in the middle of the CT scanner. The computer is used to adjust the area of ​​the sample to be scanned, and a series of continuous slice images are obtained by vertically scanning the sample.

[0047] 4) Import the sliced ​​images into the computer and perform filtering and noise reduction. Use Avizo software to reconstruct the sliced ​​images into a 3D model.

[0048] 5) Use software cutting tools to cut, preserving the sample center. The soil column image was used to remove errors caused by edge effects.

[0049] 6) Based on the grayscale difference between the pores and the matrix, the pores are extracted using a threshold segmentation method; then, the LabelAnalysis module is used to calculate and extract the pore structure parameters. Specifically, the pore structure parameters include:

[0050] Porosity:

[0051] Definition: The ratio of pore volume to total volume, reflecting the proportion of pores in the soil.

[0052] Acquisition method: The grayscale image obtained by CT scan is binarized to distinguish the pores (low-density areas) from the soil matrix (high-density areas), and then the volume ratio occupied by the pores is calculated.

[0053] Pore ​​size distribution:

[0054] Definition: The distribution of pores classified by size, usually statistically analyzed in units of pore diameter or volume.

[0055] Acquisition method: By utilizing the grayscale differences in CT scan images, pores of different sizes are extracted using image segmentation techniques, and their distribution is statistically analyzed.

[0056] Pore ​​connectivity:

[0057] Definition: The degree of connectivity between pores, which affects the transport of moisture and gas.

[0058] Acquisition method: By constructing a pore network model, analyze the connection paths and connectivity parameters between pores, such as coordination number (the number of connections between a pore and its adjacent pores).

[0059] Gap curvature:

[0060] Definition: The degree of tortuosity of the pore path affects the flow resistance of fluid in the pores.

[0061] Acquisition method: The tortuosity of the pores is calculated by analyzing the pore paths in CT scan images.

[0062] Porosity volume fraction:

[0063] Definition: The proportion of volume occupied by pores of different sizes.

[0064] Acquisition method: The pore volume fraction is calculated by statistically analyzing the volume distribution of different pore sizes in CT scan images.

[0065] Through the above steps, CT scanning technology can be used to obtain detailed information on soil pore structure, providing an important basis for the research and optimization of soil physical properties.

[0066] Step 3: Obtain the hydrothermal transfer characteristics of the soil sample.

[0067] By combining pore structure parameters obtained from CT scans, the hydrothermal transfer characteristics of soil samples were studied through numerical simulation and experiments.

[0068] In one possible embodiment, the lattice Boltzmann model can be used to study the water and heat transport characteristics of soil and obtain characteristic parameters, which can be achieved through the following specific steps:

[0069] 1) Establish a hydrothermal coupling model

[0070] Model selection: A dual distribution function is used to describe the evolution of the temperature and moisture fields respectively. The evolution of the temperature and moisture fields is simulated by two independent distribution functions, thus achieving hydrothermal coupling.

[0071] Governing equations: Based on the hydrothermal coupling governing equations of unsaturated soil, the governing equations for the temperature field and the moisture field are as follows:

[0072] Temperature field governing equations:

[0073] Moisture field governing equations:

[0074] Where T is temperature, θ is volumetric water content, α is thermal diffusivity, φ is heat source intensity, and D... θ D is the water diffusion coefficient. T is the temperature-induced moisture diffusion coefficient.

[0075] 2) Parameter settings and dimensionless transformation

[0076] Parameter settings: Set model parameters according to the physical properties of the soil (such as porosity, thermal conductivity, specific heat capacity, etc.). For example, porosity n, volumetric water content θ, thermal conductivity k, etc.

[0077] Dimensionless processing: Physical parameters are made dimensionless to simplify calculations. For example, dimensionless temperature T. d The dimensions of time F0, etc., are defined as follows:

[0078] T d =(TT) min ) / (T max -T min F0 = α × T / L 2 ;

[0079] 3) Model Discretization and Numerical Solution

[0080] Discretization: The control equations are discretized using the lattice Boltzmann method. For example, discretization is performed based on the D2Q5 model (two-dimensional five-velocity model).

[0081] Collision and Velocity Update: Based on the collision and velocity update formulas of the lattice Boltzmann method, the distribution function and macroscopic quantities (such as temperature and water content) are calculated for each time step. The specific formulas are as follows: f i (x+e i Δt, t+Δt)=[f i (x,t)-f ieq [(x,t)] / τ; where f i Let f be the distribution function. ieqLet τ be the equilibrium distribution function and τ be the relaxation time.

[0082] 4) Model Validation

[0083] Verification method: The accuracy of the model is verified by comparing it with existing analytical or numerical solutions (such as the SPH solution).

[0084] Case study: For example, simulating a water-thermal coupling problem in a semi-infinite space, comparing the temperature and moisture field distributions under different coupling modes (unidirectional coupling and bidirectional coupling).

[0085] 5) Parameter Acquisition and Analysis

[0086] Characteristic parameter calculation: Obtain the water and heat transfer characteristic parameters of the soil through simulation results.

[0087] Among them, the hydrothermal transport characteristic parameters include moisture transport characteristic parameters, heat transport characteristic parameters, and hydrothermal coupling characteristic parameters;

[0088] Moisture transport characteristic parameters include:

[0089] Saturated hydraulic conductivity:

[0090] Definition: The ability of soil to transport water through soil pores under saturated conditions, usually expressed as the amount of water passing through a unit area per unit time (e.g., cm / s or mm / h).

[0091] Function: It determines the rate at which water infiltrates the soil after irrigation or rainfall.

[0092] Unsaturated hydraulic conductivity:

[0093] Definition: The ability of soil to transport water through soil pores in an unsaturated state is related to soil moisture content (θ).

[0094] Function: It affects the retention and transport of water in the soil, especially in arid and semi-arid regions.

[0095] Soil water potential:

[0096] Definition: The free energy state of water in soil, reflecting the energy state of water in soil, and the unit is Pascal (Pa) or bar (bar).

[0097] Composition: matrix potential (ψm, related to soil pore structure), gravitational potential (ψg, related to soil depth), and pressure potential (ψp, related to soil pressure).

[0098] Function: Determines the direction of water movement in the soil, moving from areas of high water potential to areas of low water potential.

[0099] Wilting coefficient:

[0100] It is used to reflect the soil's ability to retain moisture.

[0101] Function: Affects the water supply to plants and the soil's water retention capacity.

[0102] Heat transfer characteristic parameters include:

[0103] Thermal conductivity:

[0104] Definition: The ability of soil to transfer heat through pores and solid particles, measured in W / (m·K).

[0105] Function: It determines the rate of heat transfer in the soil during diurnal temperature variations or seasonal changes.

[0106] Thermal diffusivity:

[0107] Definition: The rate at which heat diffuses through soil is the ratio of thermal conductivity to heat capacity, measured in meters per second (m). 2 / s.

[0108] Function: It reflects the soil's response speed to changes in heat. Soils with high thermal diffusivity transfer heat faster.

[0109] Soil temperature gradient:

[0110] Definition: The spatial rate of change of temperature in soil, usually expressed as temperature difference divided by distance (K / m).

[0111] Function: Affects the vertical and horizontal direction of heat transfer in the soil.

[0112] Heat capacity:

[0113] Definition: The amount of heat required to raise the temperature of a unit volume of soil by 1 degree Celsius, expressed in J / (m³). 3 ·K).

[0114] Function: It affects the rate of temperature change in soil; soils with higher heat capacity experience slower temperature changes.

[0115] Hydrothermal coupling characteristic parameters include:

[0116] Phase transition heat:

[0117] Definition: The heat absorbed or released by soil moisture during phase change processes (such as liquid water evaporating into water vapor or ice melting into liquid water).

[0118] Function: It affects soil temperature changes and moisture balance.

[0119] Step 4: Identify the types of pore structure parameters that have a significant impact on hydrothermal transfer parameters.

[0120] Specifically: A LightGBM model is constructed using pore structure parameters as input features and water and heat transfer characteristics as prediction parameters. The model is initialized using the LightGBM library, the basic parameters of the LightGBM model are set, the hyperparameters of the model are tuned using grid search or random search methods, and the LightGBM model is trained until its performance indicators reach the expected values. The importance score of each input feature is obtained through the trained LightGBM model to identify the types of pore structure parameters that have a significant impact on water and heat transfer parameters.

[0121] The specific process includes the following steps:

[0122] 1) Data preparation and preprocessing

[0123] Data collection: Information on the pore structure of the soil is obtained through CT scanning technology.

[0124] Feature extraction: Pore structure feature parameters are extracted from CT scan images and used as input features. Simultaneously, corresponding hydrothermal transport characteristic parameters are obtained through experiments or numerical simulations.

[0125] Data cleaning: Remove outliers and missing values ​​to ensure data quality and integrity.

[0126] 2) Data partitioning

[0127] Split the dataset into training and test sets: Randomly divide the dataset into training and test sets, typically in a ratio of 70% (training set) and 30% (test set), to ensure the model's generalization ability.

[0128] 3) Feature Engineering

[0129] Feature selection: Based on the correlation analysis between pore structure characteristics and hydrothermal transport characteristics, potentially important features were initially screened.

[0130] Feature encoding: Encoding non-numerical features.

[0131] 4) LightGBM Model Construction

[0132] Model initialization: Initialize the model using the LightGBM library, setting basic parameters such as learning rate, tree depth, and number of leaf nodes.

[0133] Parameter tuning: The hyperparameters of the model are tuned using methods such as grid search or random search to improve the model's performance.

[0134] 5) Model training and validation

[0135] Training the model: The LightGBM model is trained using the training set data.

[0136] Model validation: The performance of the model is validated using test set data. Evaluation metrics include mean squared error (MSE) and coefficient of determination (R²). 2 )wait.

[0137] 6) Feature Importance Analysis

[0138] After training, the importance score of each feature is obtained through the feature_importances_ attribute of the LightGBM model.

[0139] Sorting and Visualization: Sort features by importance score and display them using visualization tools such as bar charts.

[0140] 7) Identification of key pore structure parameters

[0141] Identifying key features: Based on feature importance scores, identify the types of pore structure parameters that have the greatest impact on hydrothermal transport characteristics. The specific process includes the following steps:

[0142] The importance score of the input features to the predicted parameters is obtained through the trained LightGBM model, denoted as... This is represented as the importance score of the i-th input feature to the j-th prediction parameter;

[0143] Define the parameter filtering criteria, including:

[0144] Selection criterion 1: The importance score of any i-th input feature to the j-th prediction parameter. Not less than the first screening threshold;

[0145] Filtering condition 2: For any j-th prediction parameter, the sum of the importance scores of the selected input features is not less than the second filtering threshold;

[0146] Filtering condition 3: For the i-th input feature, the sum of its importance scores to all predicted parameters is not less than the third filtering threshold;

[0147] Based on the above screening criteria, a screening model is constructed for pore structure parameter types that have a significant impact on hydrothermal transport parameters. The expression of the screening model is as follows:

[0148]

[0149] Where n is the number of prediction parameter categories; m is the number of selected input feature categories; k1 is the first screening threshold; k2 is the second screening threshold; k3 is the third screening threshold, where...

[0150] The range of the first screening threshold is k1∈(0,0.1];

[0151] The second screening threshold ranges from k2 to [0.9, 1].

[0152] The range of the third screening threshold is k3∈[0.1,0.3].

[0153] The screening model is used to select the input features that meet the criteria as parameter types that have a significant impact on hydrothermal transfer parameters.

[0154] Step 5: Using the type of pore structure parameter with significant influence as the optimization object, the optimal water and heat transfer characteristics of the soil sample as the optimization objective, and the range of values ​​of the optimization object as the constraint, construct a soil structure optimization model for the target area, use the soil structure optimization model to obtain the optimal value of the optimization object, and then form a soil structure optimization strategy.

[0155] Here is a possible model expression:

[0156] Optimization goal:

[0157] Objective 1: To maximize the thermal conductivity (k) of the soil in order to improve the efficiency of heat transfer in the soil.

[0158] Objective 2: To maximize the unsaturated hydraulic conductivity (D) of the soil to improve the efficiency of soil water transport.

[0159] Optimization target:

[0160] Porosity (n): The proportion of the volume occupied by pores in soil.

[0161] Pore ​​size distribution (d): The diameter distribution of pores in the soil.

[0162] Pore ​​connectivity (c): The degree of connectivity between pores in the soil.

[0163] Constraints

[0164] Porosity range: n min ≤n≤n max ;n min n max These represent the lower and upper limits of porosity, respectively.

[0165] Pore ​​size distribution range: d min ≤d≤d max ;d min d max These represent the lower and upper limits of the pore size distribution, respectively.

[0166] Pore ​​connectivity range: c min ≤c≤c max c min c maxThese represent the lower and upper limits of pore connectivity, respectively.

[0167] Multi-objective optimization model expression

[0168] Max f1(n,d,c)=k(n,d,c)

[0169] Max f2(n,d,c)=D(n,d,c)

[0170] Subject to

[0171] n min ≤n≤n max ;

[0172] d min ≤d≤d max ;

[0173] c min ≤c≤c max ;

[0174] Where k(n,d,c) and D(n,d,c) represent thermal conductivity and unsaturated hydroconductivity as functions of porosity, pore size distribution, and pore connectivity, respectively.

[0175] By solving the above multi-objective optimization model, we can find the optimal set of pore structure parameters that enable the soil to have the best water and heat transfer characteristics under given constraints. These optimal solutions can provide a scientific basis and guidance for soil structure optimization.

[0176] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing soil structure based on CT scanning and hydrothermal transfer characteristics analysis, characterized in that, include: Soil samples from the target area were obtained and processed. The processed soil samples were scanned using an industrial CT scanner. The images obtained from the CT scans were preprocessed. The CT images were segmented using image processing software. The pore and solid phases were distinguished based on the difference in gray values, and the pore structure parameters were extracted. Combined with the pore structure parameters obtained from the CT scans, the hydrothermal transfer characteristics of the soil samples were studied through numerical simulation and experiments to obtain the hydrothermal transfer characteristic parameters of the soil samples. Using the pore structure parameters as input features and the water and heat transfer characteristic parameters as prediction parameters, a LightGBM model is constructed. The model is initialized using the LightGBM library, and the basic parameters of the LightGBM model are set. The hyperparameters of the model are tuned using grid search or random search methods. The LightGBM model is trained until its performance index reaches the expected value. The importance score of each input feature is obtained through the trained LightGBM model to identify the types of pore structure parameters that have a significant impact on the water and heat transfer parameters. Using the pore structure parameter type with significant influence as the optimization object, the optimal hydrothermal transfer characteristics of the soil sample as the optimization objective, and the value range of the optimization object as the constraint, a soil structure optimization model for the target area is constructed. The optimal value of the optimization object is obtained using the soil structure optimization model, thereby forming a soil structure optimization strategy.

2. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 1, characterized in that, The pore structure parameters include one or more of the following: porosity, pore size distribution, pore connectivity, pore tortuosity, and pore volume fraction.

3. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 1, characterized in that, The hydrothermal transport characteristic parameters include moisture transport characteristic parameters, heat transport characteristic parameters, and hydrothermal coupling characteristic parameters, wherein: The water transport characteristic parameters include one or more of the following: saturated hydraulic conductivity, unsaturated hydraulic conductivity, soil water potential, and wilting coefficient. The heat transfer characteristic parameters include one or more of the following: thermal conductivity, thermal diffusivity, soil temperature gradient, and heat capacity. The hydrothermal coupling characteristic parameters include phase change heat.

4. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 1, characterized in that, The basic parameters of the LightGBM model include the learning rate, the depth of the tree, and the number of leaf nodes; the performance metrics of the LightGBM model include either the mean squared error or the coefficient of determination.

5. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 1, characterized in that, The process for identifying the type of pore structure parameter that has a significant impact on the hydrothermal transfer parameters includes the following steps: The importance score of the input feature to the prediction parameter is obtained through the trained LightGBM model, denoted as . It is represented as the importance score of the input feature described in the i-th item to the prediction parameter described in the j-th item; Define the parameter filtering criteria, including: Filtering criterion 1: The importance score of any input feature described in the i-th item to the prediction parameter described in the j-th item. Not less than the first screening threshold; Filtering condition 2: For any j-th prediction parameter, the sum of the importance scores of the selected input features is not less than the second filtering threshold; Filtering condition 3: For the input feature described in item i, the sum of its importance scores to all the predicted parameters is not less than the third filtering threshold; Based on the screening conditions described above, a screening model is constructed for the pore structure parameter types that have a significant impact on the hydrothermal transfer parameters. The screening model is then used to select the input features that meet the criteria as parameter types that have a significant impact on the hydrothermal transfer parameters.

6. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 5, characterized in that, The expression for the screening model is: Where n is the number of types of prediction parameters; m is the number of types of selected input features; k1 is the first screening threshold; k2 is the second screening threshold; and k3 is the third screening threshold.

7. The soil structure optimization method based on CT scanning and hydrothermal transfer characteristic analysis according to claim 6, characterized in that, The first screening threshold ranges from k1 to (0, 0.1). The second screening threshold ranges from k2 to [0.9, 1]. The value range of the third screening threshold is k3∈[0.1,0.3].

Images