A soil-water characteristic relationship prediction method based on big data analysis

Abstract

This invention provides a method for predicting soil-water characteristic relationships based on big data analysis. First, a descriptive model of soil pore size distribution is established based on fractal geometry. Then, a capillary suction model considering pore fractal characteristics is established by combining capillary theory. This theoretical model enables rapid and reliable prediction of soil-water characteristic relationships under the condition of known soil pore fractal structure. The model prediction method utilizes big data analysis, collecting measured soil-water characteristic curves from a large number of soil samples. It then uses the theoretical model to infer the fractal characteristics of the soil pore structure, establishing a correspondence between soil apparent physical indicators and pore fractal parameters. This method enables rapid prediction of the pore fractal structure of any soil sample based on its apparent physical indicators. Combining this theoretical model and prediction method, the limitations of traditional methods, such as long testing times and high costs, are overcome, achieving rapid and reliable prediction of soil-water characteristic curves.

Description

Technical Field

[0001] This invention relates to the fields of soil science, data analysis and prediction technology, specifically a method for predicting soil water characteristic curves based on big data analysis. Background Technology

[0002] The soil-water characteristic curve describes the relationship between soil water saturation and soil water suction. This relationship is fundamental to research on slope seepage and control in hydraulic engineering, as well as research on soil hydrodynamic properties, soil water regulation and utilization, and soil improvement in agricultural engineering. Therefore, the measurement and prediction of soil-water characteristic curves are of great importance and practical significance in the field of geotechnical engineering.

[0003] Traditional methods for determining soil-water characteristic curves include direct measurement methods and indirect prediction methods. Direct measurement methods typically rely on laboratory tests, such as pressure plate methods, centrifugation methods, and axis translation techniques. These methods are time-consuming, costly, and highly dependent on testing equipment and operator skills, introducing numerous uncertainties. Indirect prediction methods, on the other hand, use mathematical models to predict soil-water characteristic curves, including empirical formulas, soil transformation functions, and physical empirical models. Indirect prediction methods overcome the time-consuming, labor-intensive, and uneconomical nature of experimental curve determination, becoming a simpler, faster, and more effective approach. However, the parameters in these mathematical models often lack clearly defined physical meanings.

[0004] Soil-water characteristic curves are essentially a macroscopic reflection of soil pore structure. Therefore, some scholars have attempted to establish mathematical models based on soil pore structure to predict soil-water characteristic curves. However, pore structure is affected by many factors such as compaction and particle size distribution, and accurate prediction remains a challenge to date. Summary of the Invention

[0005] To address the above issues, this invention provides a method, apparatus, and device for predicting soil-water characteristic curves based on big data analysis, enabling rapid and accurate prediction. The method consists of two parts: first, a soil-water characteristic curve prediction model based on pore structure distribution is established; then, the correspondence between pore size distribution and indicators such as dry density and particle size distribution is obtained through big data statistics; finally, rapid and accurate prediction of soil-water characteristic curves based on apparent physical indicators of the soil (such as dry density and particle size distribution) is achieved. The technical solution of this invention is as follows: A method for predicting soil-water characteristic curves based on big data analysis includes the following steps: Step 1: Establish a soil-water characteristic curve prediction model based on pore structure distribution. The Menger sponge model, a commonly used porous fractal model for soil and rock media, was adopted to establish a three-dimensional fractal description of the soil pore structure. The specific construction method includes: ① First, assuming the soil has an initial side length (the size of the computational domain side length) of... L ① The cube; ② The cube is made of m After dividing into equal parts, there exists m 3 A number of cubes are randomly removed, leaving... j ③ Continue dividing and eliminating all remaining cubes using the same method. At this point, the number of remaining small cubes increases, and their size decreases. ④ When the operation is in progress... k The number of cubes at this time is j k The side length is R = L / m k For determining the soil sample, parameters j , k , m All are constants, and their fractal dimension is D = lg( j ) / lg( m ).

[0006] As an improvement, the total volume of soil in the three-dimensional sponge model V t From the solid volume of soil V s With total pore volume V p Composition, total volume of soil V t Solid volume of soil V s With total pore volume V p They are respectively: (1); In the formula, R Let L be the minimum pore radius of the soil sample, and L be the side length of the computational domain. D It is the fractal dimension.

[0007] During soil dehydration, large pores in the soil always lose water first, followed by small pores. If the pore radius at the water-air interface is... r Therefore, it can be assumed that pores larger than this diameter are entirely filled with air, while pores smaller than this diameter are entirely filled with water. Thus, pores with a diameter greater than or equal to this diameter are... r The pore volume (i.e., the volume filled with air). Vp (≥ r The percentage of total volume, expressed as (2); air-filled volume V p (≥ r The percentage of the total volume is expressed as: (3); In the formula V p (≤ r () represents the volume of water in the soil. V s For soil particles, V t The total volume of the soil is given. The volume of a substance is closely related to its mass. The physical quantities involved in the above formula are expressed as follows: (4); In the formula: m w The mass of water in the soil. m s For soil particle mass, d s This refers to the relative density of soil particles; ρ w The density of water; ρ d The dry density of the soil. w This refers to the soil's moisture content.

[0008] By combining equations (2), (3), and (4), we can obtain: (5); Taking the logarithm of both sides of the above equation, we get the soil mass moisture content. w With critical radius r The relationship is represented as (6); As an improvement, the soil pore fractal model uses an ideal capillary tube to describe the pores and the Young-Laplace equation is used to establish capillary suction. S Size and pore radius r Contact angle of solid-liquid-gas three-phase interface θ Relationship (7); In the formula: r Where is the pore radius, S Capillary suction, T s is the surface tension of the liquid.

[0009] Combining equations (6) and (7), we can obtain the result considering dry density. ρ d Capillary suction under influence S With soil mass and moisture content w Relationship (8); The prediction model includes T s , θ , L , d s , D , ρ d There are a total of 6 unknown parameters: T s The surface tension is taken as 0.073 N / m at 25℃; θ The contact angle is 0 during the dehumidification process; d s The relative density of the soil is measured using the hydrometer bottle method; dry density... ρ d Measured using the drying method.

[0010] Step 2: Establish soil pore fractal feature parameters (fractal dimension) based on BP neural network. D and L, L To calculate the domain side length, this formula establishes the relationship between pore size distribution (corresponding to the maximum pore size) and indices such as dry density and particle size distribution. The specific method includes: As an improvement, soil moisture content data from numerous previous experimental tests were collected. w ~Soil water suction S The curve data and its corresponding soil physical parameters, such as relative density, dry density, particle size distribution, etc., are collected and a database is established. As an improvement, based on the data in the database and combined with the soil fractal model, the distribution of soil pore size characteristics was plotted. The specific steps include: ① According to formula (8), the left-hand side of the formula is used to plot the distribution of soil pore size characteristics. Using the vertical axis as the ordinate, and the right-hand side as the index. ① Plot the data points using the horizontal axis as the x-axis and perform linear fitting on the data points; ② Obtain the slope based on the linear fitting results. and intercept , then it can be calculated Fractal dimensions D and L, L represents the side length of the computational domain, which corresponds to the maximum aperture in this formula.

[0011] As an improvement, different dry densities in the database are addressed. ρ d Particle size distribution dSoil samples underwent data preprocessing, including cleaning, noise reduction, and normalization, and the dataset was divided into training and testing sets.

[0012] As an improvement, the topology of the BP neural network is determined, with apparent physical parameters of the soil as input and pore fractal parameters as output, and the number of neurons in the hidden layer is determined according to empirical formulas: (9); In the formula, n This represents the number of neurons in the hidden layer. m The number of nodes in the input layer. l The number of nodes in the output layer. k ∈ (0,10).

[0013] As an improvement, initial weights and thresholds are input into the neural network for training. The error between the real data and the predicted data of the pore fractal parameters is calculated. When the error value does not meet the preset accuracy requirement, it is fed back from the output layer to the input layer layer by layer, and the connection weights and thresholds between each layer are continuously adjusted until the error reaches the set stopping condition.

[0014] As an improvement, the test set was run with optimized connection weights and thresholds to verify the prediction accuracy of the trained BP neural network.

[0015] The present invention has the following beneficial effects: 1. The model proposed in this invention is based on fractal theory to establish a description of soil pore structure, overcoming the shortcomings of traditional models where the physical meaning of parameters is unclear and the values ​​are difficult to obtain. The characteristic parameters involved in the model have clear physical meanings and can be quickly measured through simple experiments. Furthermore, the capillary suction prediction model established by combining the capillary model (Young-Laplace equation) can achieve accurate and reliable prediction of capillary suction.

[0016] 2. A large amount of measured data on soil capillary suction was collected, and a database for soil capillary suction testing was established. Furthermore, statistical analysis of the large dataset was conducted in conjunction with a soil pore structure model, and model characteristic parameters (fractal dimension) were established. D The correlation between soil pore size (maximum pore diameter) and apparent physical quantities of soil (dry density, particle size distribution, etc.) enables rapid prediction of soil pore structure. Compared with traditional methods for directly measuring soil pore structure, this prediction method is not only faster but also almost cost-free.

[0017] 3. In the statistical analysis of large amounts of data, a BP neural network intelligent algorithm was further employed to deeply mine the collected data, overcoming the difficulty of analyzing the coupled influence of multiple physical factors (dry density, particle size distribution, relative density, plasticity index, etc.) in traditional data analysis. This not only enabled the analysis of model characteristic parameters (fractal dimension) under the coupled influence of multiple physical factors, but also... D It can predict the maximum aperture and also achieve a quantitative evaluation of the importance of each factor.

[0018] 4. By combining the capillary suction prediction model and the pore structure prediction method, we can quickly and reliably predict the soil-water relationship based solely on the apparent physical parameters of the soil. Compared with the traditional direct testing method, this method can obtain a large amount of soil-water relationship data quickly, so as to provide timely guidance for engineering projects. Compared with the traditional indirect prediction method, this model and method have clear physical meaning, and the prediction results are more reliable. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method for predicting soil-water characteristic curves based on pore structure according to the present invention; Figure 2 The flowchart for establishing a database of soil fractal model parameters and apparent physical indices for this invention; Figure 3 The soil-water characteristic curve data of the collected samples are shown in the figure for this invention. Figure 4 This is a diagram showing the fractal characteristics of soil pores according to the present invention. Figure 5 This is a flowchart of the BP neural network algorithm of the present invention; Figure 6 This is a comparison chart of the actual and predicted values ​​of the fractal dimension of this invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, but the implementation of the present invention is not limited thereto. Example 1 In this invention, for ease of explanation of the model and method, the prediction of the soil-water characteristic curve of loess is used as an example, and the specific steps are as follows: Figure 1 As shown, it includes 4 steps from S1 to S4.

[0021] Step S1: Establish a database of the correspondence between soil apparent physical properties (dry density, particle size distribution) and pore structure fractal parameters. This step can be further broken down into S101~S105. For example... Figure 2-4 As shown.

[0022] Sub-step S101: Collect previously measured soil mass moisture content ~ capillary suction curve data, such as... Figure 3As shown.

[0023] Sub-step S102: Collect the corresponding soil physical parameters from step S101, such as dry density, particle size distribution, and relative density of soil particles, as shown in Table 1.

[0024] Table 1 Physical parameters of soil samples

[0025] Sub-step S103: Substitute the soil mass moisture content ~ capillary suction data into formula (8), and simultaneously use log( S ) is the horizontal axis, log[ w+1 / d s Using [] as the ordinate axis, plot the loess data points, as shown below. Figure 4 As shown.

[0026] Sub-step S104: Further perform linear fitting on the data points ( Figure 4 ), to obtain the slope of the fitted function ) and intercept Calculate the fractal dimension D and maximum aperture L (computational domain edge) (Length and size) As shown in Table 2.

[0027] Table 2 Fractal characteristic parameters of loess samples

[0028] Sub-step S105: Collect a large amount of measured soil sample data. Repeat steps S101 to S104 to establish the correspondence between the fractal dimension of each soil sample and the physical properties of the soil, as shown in Table 2: dry density ~ fractal dimension / maximum pore size. Based on the large amount of measured data collected, establish a database.

[0029] Step S2: Train and test the dataset using a BP neural network to obtain the BP neural network prediction model best suited for the dataset, using fractal dimension. D For example, this step can be broken down into S201 to S204.

[0030] Sub-step S201: Preprocess the raw data in the database, including cleaning, denoising, normalization, etc., and divide the dataset into training set and test set in a ratio of 4:1.

[0031] Sub-step S202: Using the apparent physical parameters of the soil as the input layer, including six characteristic parameters: relative density of soil particles, dry density, plasticity index, effective particle size, median particle size, and limiting particle size, and using fractal dimension as the input. D As the output layer, the range of hidden layer neurons is determined to be 3 to 12 according to formula (9).

[0032] Sub-step S203: Based on this range, run the neural network to perform trial calculations and compare the mean square error values ​​of neurons in different hidden layers. When the number of neurons in the hidden layer is 11, the mean square error value of the BP neural network is the smallest. Therefore, the optimal number of neurons in the hidden layer can be determined to be 11.

[0033] Sub-step S204: By determining the coefficient R 2 The mean absolute error (MAE) and root mean square error (RMSE) are used to evaluate the prediction accuracy of the model, and the optimal model is selected.

[0034] Step S3: Obtain the apparent physical properties of the soil to be tested through simple experiments. For example, the basic physical properties of a loess sample from a certain area are as follows: Table 3 Basic physical properties of soil samples

[0035] Step S4: Using the tested apparent physical properties of the soil as input, the BP neural network model trained in step S2 is used to predict the fractal parameters of the soil's pore structure. D , L The fractal dimension of the soil sample was predicted to be D=2.8152 and the maximum pore size was L=0.3524.

[0036] Step S5: Calculate the predicted fractal model parameters. D , L Substituting into formula (8), the soil-water characteristic curve of the soil sample to be tested can be established. Figure 6 It can be seen that for the 43 soil samples collected, the trend of fractal dimension change predicted by the BP neural network basically conforms to the general trend of the actual fractal dimension curve, and the predicted value of the fractal dimension of most samples almost coincides with the true value. Therefore, the BP neural network can predict the fractal dimension of this dataset very well.

[0037] The technical solutions of the present invention have been explained through the above embodiments, but the present invention is not limited to the above embodiments, that is, it does not mean that the present invention must rely on the above specific embodiments to be implemented. Any improvements made by those skilled in the art based on the present invention, or equivalent substitutions for the materials selected in the present invention, fall within the scope of patent protection.

Claims

1. A method for predicting the soil-water characteristic relationship based on big data analysis, characterized in that, Includes the following steps: Step 1: Establish a soil-water characteristic curve prediction model based on pore structure distribution. The Menger three-dimensional sponge model, a commonly used porous fractal model for soil and rock media, was adopted to establish a three-dimensional fractal description of soil pore structure. Furthermore, a formula for predicting capillary suction based on soil pore structure was proposed by combining the Yang-Laplace equation. Capillary suction S With soil moisture content w The relationship is as follows: （8）； The prediction model includes T s , θ , L , d s , D , ρ d There are a total of 6 unknown parameters: T s The surface tension is taken as 0.073 N / m at 25℃; θ The contact angle is 0 during the dehumidification process; d s The relative density of the soil is measured using the hydrometer bottle method; dry density... ρ d Measured according to the drying method; L is the side length of the calculation domain, which corresponds to the maximum aperture in this formula; D It is the fractal dimension; w S represents the soil mass moisture content; S represents capillary suction. Step 2: Establish the correspondence between soil pore fractal characteristic parameters and soil apparent physical indices based on BP neural network. Step 3: Test the apparent physical properties of the soil to be tested, use the relationship obtained in Step 2 to predict the soil pore fractal parameters, and further combine with Step 1 to predict the soil-water characteristic curve. Step 2 specifically includes: collecting soil moisture content data from experimental tests. w ~Soil water suction S Curve data and their corresponding apparent physical parameters of the soil, including relative density, dry density, particle size distribution, or others, and establish their database; soil pore fractal characteristic parameters, including fractal dimension. D and maximum aperture L The apparent physical properties of soil include relative density, dry density, particle size distribution, or others; The specific steps for establishing the correspondence between soil pore fractal characteristic parameters and soil apparent physical properties based on a BP neural network include: 2.1 According to formula (8), the terms on the left side of the formula are... Using the vertical axis as the ordinate, and the right-hand side as the index. Plot the data points using the horizontal axis as the x-axis and perform linear fitting on the data points; 2.2 Obtain the slope based on the linear fitting results and intercept Calculate the fractal dimension D And the side length L of the computational domain, where L corresponds to the maximum aperture in this formula; 2.3 For different dry densities in the database ρ d Particle size distribution d Soil samples underwent data preprocessing, including cleaning, noise reduction, and normalization, and the dataset was divided into training and testing sets. 2.4 Determine the topology of the BP neural network, using the apparent physical parameters of the soil as input and the pore fractal parameters as output, and determine the number of neurons in the hidden layer based on empirical formulas; （9）； In the formula, n This represents the number of neurons in the hidden layer. m The number of nodes in the input layer. l The number of nodes in the output layer. k ∈ (0,10); 2.5 Input the initial weights and thresholds into the neural network to train it, calculate the error between the real data and the predicted data of the pore fractal parameters, and when the error value does not reach the preset accuracy requirement, feed back from the output layer to the input layer layer by layer, continuously adjust the connection weights and thresholds between each layer until the error reaches the set stopping condition. 2.6 Run the test set with the optimized connection weights and thresholds to verify the prediction accuracy of the trained BP neural network.

2. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 1, characterized in that, The specific construction method of the three-dimensional sponge model in step 1 includes: 1.1 Let the soil mass be the initial side length corresponding to the side length of the computational domain. L A cube; 1.2 Place the cube with m After dividing into equal parts, there exists m 3 A number of cubes are randomly removed, leaving... j A small cube; 1.3 Continue to divide and eliminate all remaining cubes using the same method. At this point, the number of remaining small cubes increases and their size decreases. 1.4 When the operation is performed k The number of cubes at this time is j k The side length is R = L / m k For determining the soil sample, parameters j , k , m All are constants, and their fractal dimension is D = lg( j ) / lg( m ).

3. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 2, characterized in that, Total soil volume in a 3D sponge model V t From the solid volume of soil V s With total pore volume V p composition: （1）； In the formula, R is the minimum pore radius of the soil sample; L is the side length of the computational domain, which corresponds to the initial side length here. D It is the fractal dimension.

4. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 2, characterized in that, In the three-dimensional sponge model, the pore radius at the water-air interface is assumed to be... r Pores larger than this diameter are entirely filled with air, while pores smaller than this diameter are entirely filled with water. Pores with a diameter greater than or equal to this diameter are... r The pore volume is the volume filled with air. V p (≥ r The percentage of the total volume is expressed as: （2）； air-filled volume V p (≥ r The percentage of the total volume is expressed as: （3）； In the formula V p (≤ r () represents the volume of water in the soil. V s For soil particles, V t The total volume of the soil is given by the formula. The volume of a substance is closely related to its mass. The physical quantities involved in the formula are expressed as follows: （4）； In the formula m w The mass of water in the soil. m s For soil particle mass, d s The relative density of soil particles; ρ w The density of water; ρ d The dry density of the soil. w This refers to the soil's moisture content.

5. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 4, characterized in that, Combining equations (2), (3), and (4), we obtain the following relationship: （5）。 6. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 5, characterized in that, Soil mass moisture content w With critical radius r The relationship is that the fractal model of soil pore structure is: （6）。 7. The method for predicting soil-water characteristic relationships based on big data analysis according to claim 6, characterized in that, In the soil pore fractal model, the Young-Laplace equation is used to establish capillary suction. S Size and pore radius r Contact angle of solid-liquid-gas three-phase interface θ Relationship: （7）； In the formula: r Where is the pore radius, S Capillary suction, T s The surface tension of the liquid; Substituting equation (7) into equation (6) yields the capillary suction force. S With soil moisture content w The relationship between them (8).

Images