Soil compaction state classification method based on deep learning network

Abstract

The application discloses a soil compaction state grade classification method based on a deep learning network. The method comprises the following steps: obtaining soil compaction state grade classification; arranging a geophone in a target area, picking up a dispersion curve based on a surface wave exploration method, inputting the picked dispersion curve into a soil compaction state grade classification prediction model, and obtaining the soil compaction state grade classification of the target road section; and the establishment method of the soil compaction state grade classification prediction model comprises the following steps: obtaining training samples, the training data being a dispersion curve corresponding to a constructed stratum structure, soil quality and a soil compaction state thereof; inputting the training samples into a deep learning network model, and obtaining a soil compaction state grade classification model of the target area after training. The method provided by the application is high in efficiency and can effectively avoid errors caused by individual differences, so that the soil compaction state grade of a target section can be accurately and reliably classified quickly.

Description

Technical Field

[0001] The present invention relates to the field of geophysical exploration technology, and in particular to a method for determining soil density classification. Background Art

[0002] Existing surface wave exploration technology is widely used in geophysical exploration and engineering surveys. By analyzing the propagation characteristics of surface waves, the shear wave velocity structure of the underground medium can be obtained, thereby inferring the physical and mechanical properties of the soil.

[0003] Deep learning network technology, as a cutting-edge technology in the field of artificial intelligence, has powerful data processing and pattern recognition capabilities. Deep learning networks are applied to the analysis of geophysical exploration data, especially for the assessment of soil compaction.

[0004] The present invention aims to overcome the shortcomings of existing technologies by providing a method for determining soil compaction classification. This method combines surface wave exploration technology with a deep learning network model, utilizing surface wave dispersion characteristics and information about formation velocity and density to construct a soil compaction assessment model. This method enables efficient and accurate soil compaction classification, providing a scientific basis for engineering design and geological disaster prevention. Summary of the Invention

[0005] By combining the efficient data acquisition capabilities of surface wave exploration technology with the powerful learning capabilities of deep learning networks, this method overcomes the inability of traditional methods to accurately assess soil compaction under complex geological conditions. This method enables the classification of soil compaction.

[0006] A method for classifying soil compaction levels is as follows:

[0007] Step 1: Through preliminary research on the soil types and density standards of different soil types in the target area and its surroundings, Soild = [Soild1, Soild2,…, SoildN], and the corresponding soil density state standard at the corresponding Vs speed is [Com1, Com2,…, Com3].

[0008] Furthermore, the construction standard in step 1 is as follows: Soild1: its soil compactness state is divided into slightly loose (Com1) and velocity Vs range is [v 11 ,v 12 ]; Medium density (Com2) speed Vs range is [v 13 ,v 14 ]; the speed Vs of compaction (Com3) range is [v 15 ,v 16 ]; Soil N:The soil compaction state is divided into slightly loose (Com1) and the velocity Vs range is [v N1 ,v N2 ]; Medium density (Com2) speed Vs range is [v N3 ,v N4 ]; the speed Vs of compaction (Com3) range is [v N5 ,v N6 ];

[0009] Step 2: According to the underground geological structure of the exploration area, determine the physical properties of each layer and obtain the detection depth H of the target area; divide the layers into layers according to D (m), N = H / D, and obtain N layers; for the velocity and soil quality of each layer, the Monte Carlo method can be used to randomly generate N underground structure models; including the longitudinal wave velocity Vp, shear wave velocity Vs, density ρ and thickness h, and establish a layered model.

[0010] Step 3: Obtain the theoretical dispersion curve f n , where in each layer, it is assumed that the surface wave propagates horizontally and its displacement is in the form of a plane wave, satisfying the elastic dynamics equation:

[0011]

[0012] Where u represents the particle displacement vector, ρ is the density, and λ and μ are the Lame constants.

[0013] Furthermore, in step 3, for Rayleigh waves, the following dispersion equation is derived by combining the boundary condition (free surface stress is zero) and the interlayer continuity condition (displacement and stress are continuous):

[0014] D(ω, k)=0

[0015] Where ω is the angular frequency, k is the wave number, c = ω / k, and c is the phase velocity.

[0016] Furthermore, in step 3, for the Love wave, when only the horizontal displacement component is considered, the dispersion equation can be expressed as:

[0017]

[0018] in:

[0019]

[0020] Furthermore, in step 3, the dispersion equation is numerically solved to obtain f n ,;

[0021] (1) Select the appropriate frequency range ω

[0022] (2) According to the formation model parameters, the initial phase velocity c is set and the corresponding wave number k = ω / c is calculated.

[0023] (3) Use the numerical iterative method to solve the dispersion equation D(ω,k)=0 to obtain the phase velocity c at a specific frequency.

[0024] (4) Repeat (1), (2), and (3) to obtain the corresponding relationship between frequency ω and phase velocity c, that is, the corresponding theoretical dispersion curve f n =[(ω1,c1),(ω2,c2),…,(ωn,cn)], where the phase velocity c is Vs.

[0025] Step 4: Repeat step 3 according to the different geological model structures constructed to obtain N theoretical dispersion curves F = [f1, f2, f3, ..., f n ].

[0026] Step 5: Determine Soild=[Soild1,Soild2,…,SoildN] and Vs=[[v 11 ,v 12 ],[v 21 ,v ＝22 ],…,[v N1 ,v N2 ]], the corresponding soil compaction state Com=[Com1, Com2,…, Com3] is determined by Soild and Vs;

[0027] Step 6: Construct the training set [F, Soild], the corresponding label is [Com].

[0028] Step 7: Establish a soil compaction level prediction model. The prediction model includes a shared feature layer and a soil embedding layer. The corresponding K layers establish K branches. Each branch contains a classifier. The classifier is composed of a fully connected layer.

[0029] Furthermore, in step 7, the shared feature layer is constructed as follows: each layer includes a one-dimensional convolution layer; the convolution layer is followed by a batch normalization layer; the batch normalization layer is fed into an activation function; the output after the activation function is fed into a regularization layer; the regularization layer is fed into a pooling layer;

[0030] Furthermore, in step 7, the embedding layer and the fully connected classification layer in the model are constructed as follows: each layer of soil quality enters the embedding layer; the shared feature layer and the embedding layer data of the corresponding layer are spliced; the spliced ​​data is sent to the fully connected layer of the corresponding branch to output the corresponding soil density state classification;

[0031] Step eight: the corresponding F is sent into the feature extraction layer; the different layer thickness of the soil solid is sent into the embedded layer respectively; the shared feature extraction layer and the embedded layer are spliced to obtain the corresponding branch of different soil layer depth; the branch is a full connection layer, and the corresponding output is the corresponding soil layer corresponding to the soil compaction state [Com];

[0032] Step nine: build the above model structure, and train the training set model to obtain a soil compaction state grade prediction model meeting the requirements.

[0033] Further, in the step nine, the evaluation criteria are determined by the loss function Loss and the accuracy Accuracy.

[0034] Further, in the step nine, the training process includes a training set and a validation set; the validation set is a part of the training set; wherein the loss function sizes of the training set and the validation set are Loss1 and Loss2; the accuracies are Acc1 and Acc2.

[0035] Further, in the training process of the step nine, the Loss function and the Acc need to reach certain threshold values ∈1 and ∈2, which are determined according to the training set; at the same time, the values of (Loss1-Loss2) and (Acc1-Acc2) also need to reach certain threshold values ∈3 and ∈4, which are determined according to the training set.

[0036] Step ten: arranging geophones in a target area, and analyzing and processing geophone data to obtain a dispersion energy graph E(ω, c);

[0037] Further, in the step ten, the analysis and processing of the geophone data include the following calculation methods:

[0038] A surface wave seismic record u(x, t), that is, the collected geophone data; the corresponding Fourier transform can be expressed as u(x, ω), which is composed of an amplitude term A(x, ω) and a phase term P(x, ω), that is:

[0039] u(x, ω) = A(x, ω)P(x, ω) = A(x, ω)exp(-iωt)

[0040] Wherein, ω is the circular frequency, i is the imaginary unit, the amplitude term A(x, ω) mainly reflects the influence of spherical diffusion on the amplitude and the attenuation characteristics of the medium, and the phase term P(x, ω) preserves the travel time information or phase information of the wave; the negative sign in the phase term indicates that the source is excited on the left side.

[0041] Further, in the step ten, the analysis and processing of the geophone data include the following calculation methods: under the far-field approximation condition, the diffusion mode of the wave can be regarded as plane diffusion, that is:

[0042] P(x,w)=exp(-iwt)=exp(-iwx / c w )

[0043] Where x is the distance from the source to the receiver, c w is the phase velocity corresponding to the frequency.

[0044] The surface wave dispersion energy E(ω, c) can be extracted by scanning the spectrum u(x, ω) of the seismic record with another e index. For a surface wave seismic record with N channels, the extraction formula can be expressed as:

[0045]

[0046] Step 11: Obtain the dispersion curve of the target area. The dispersion curve is obtained as follows: Pick the maximum value of the velocity at the corresponding frequency in the dispersion energy diagram E(ω, c) to obtain the dispersion curve f actual ;

[0047] Step 12: Send the dispersion curve of the target area into the trained soil density state level prediction model to obtain the corresponding prediction of the soil density state level of the target area. predict =[Com p1 ,Com p2 ,…,Com pn ]. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is the soil density state level prediction model diagram;

[0049] Figure 2 is the shared feature layer graph;

[0050] Figure 3 Graphs for the embedding layer and the fully connected classification layer;

[0051] Figure 4 The loss function and accuracy graph of the training set and validation set;

[0052] Figure 5 The prediction of the target area and the distribution map of soil density level of its actual underground structure;

[0053] Figure 6 It is the soil density standard for different soil types;

[0054] Figure 7 Flowchart for establishing a soil density classification prediction model DETAILED DESCRIPTION

[0055] The present invention is further described in detail below in conjunction with the implementation process of the specific method. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0056] For a road section with a depth of 30m and a length of 30m, the implementation method is as follows:

[0057] 1) According to the national and local standards, the soil types and density standards of the target area and its surrounding areas are established; Soild1: The soil state is divided into slightly loose velocity range [v 11 ,v 12 ]; Medium density speed range [v 13 ,v 14 ]; the speed range of density is [v 15 ,v 16 ]; SoildN: The soil state is divided into slightly loose velocity range [v N1 ,v N2 ]; Medium density speed range [v N3 ,v N4 ]; the speed range of density is [v N5 ,v N6 ], see Figure 6 ;

[0058] 2) Construct a training set for the soil compaction state prediction model, 100,000 theoretical dispersion curves F = [f1, f2, f3, ..., f n ].

[0059] 3) According to the generated underground structure model, determine Soild=[Soild1,Soild2,…,SoildN] and Vs=[[v 11 ,v 12 ],[v 21 ,v ＝22 ],…,[v N1 ,v N2 ]], the corresponding soil compaction state Com=[Com1, Com2,…, Com3] is determined by the corresponding standard;

[0060] 4) Construct the training set (F, Soild), the corresponding label is (Com);

[0061] 5) Construct a soil compaction state prediction model, see Figure 1 、 Figure 2 、 Figure 3 ; Description of the model as a whole and its shared feature layer, embedding layer and fully connected classification layer respectively.

[0062] 6) The model is trained to obtain a soil density state prediction model that meets the requirements; the loss function sizes of the training set and the validation set are Loss1 and Loss2; the accuracy rates are Acc1 and Acc2; the respective Loss functions and Acc need to reach certain thresholds ∈1 and ∈2, depending on the training set; at the same time, the values ​​of (Loss1-Loss2) and (Acc1-Acc2) also need to reach certain thresholds ∈3 and ∈4, depending on the training situation, see for details Figure 4 .

[0063] 7) Data acquisition and analysis are performed on the target area using the surface wave exploration method to obtain the dispersion curve. For surface wave seismic records with N channels, the extraction formula can be expressed as:

[0064]

[0065] For the above dispersion energy diagram, the maximum value of the velocity at the corresponding frequency is picked to obtain the dispersion curve.

[0066] 8) Using the dispersion curve obtained in the target area, the dispersion curve is fed into the trained soil density state level prediction model to obtain the soil density state level classification of the target area;

[0067] In summary, the present invention uses a neural network to establish a soil density level prediction model to predict and grade the density at different depths underground in the target area. Figure 5 ;

[0068] The above description is only a preferred example for implementing the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for classifying soil compaction levels, for use in grading soil compaction levels, characterized in that: A soil compaction level prediction model is established through a neural network to predict and classify the compaction level at different depths in the target area, including the following steps: Step 1: Through preliminary research on the soil types and density standards of the target area and its surroundings, Soild=[Soild1, Soild2,…, Soildn], the corresponding soil density state standard at the corresponding Vs speed is [Com1, Com2,…, Com n ]; Step 2: Obtain the target area detection depth H; divide the layers into layers according to D, N = H / D, and obtain N layers; randomly generate N underground structure models based on the velocity and soil quality of each layer using the Monte Carlo method; Step 3: According to the generated underground structure model, obtain the theoretical dispersion curve of the underground structure model; Step 4: Based on the generated underground structure model, repeatedly calculate step 3 to obtain N theoretical dispersion curves F = [f1, f2, f3, ..., f N ]; Step 5: Determine Soild_ based on the generated underground structure model 实际 =[Soild1_ 实际 ,Soild2_ 实际 ,…,SoildN_ 实际 ] and Vs_ 实际 =[[v 11 ,v 12 ],[v 21 ,v 22 ],…,[v N1 ,v N2 ]], the corresponding soil compaction state Com_ is determined by Soild and Vs 实际 =[Com1_ 实际 ,Com2_ 实际 ,…,Com N _ 实际 ]; Step 6: Construct training set (F, Soild_ 实际 ), the corresponding label is Com_ 实际 ; Step 7: Establish a soil compaction level prediction model. The prediction model includes a shared feature layer and a soil embedding layer. N branches are established for each of the N layers. Each branch contains a classifier. The classifier is composed of a fully connected layer. Step 8: Send F to the feature extraction layer accordingly; Soild_ 实际 The corresponding features are sent to the embedding layer respectively; after the shared feature extraction layer and the embedding layer are spliced ​​together, the corresponding branches of different soil depths will be obtained; Its branch is a fully connected layer, and the corresponding output is the corresponding soil layer and the corresponding soil density state Com_ 实际 ; Step 9: Use the training set to train the constructed model to obtain a soil density state level prediction model that meets the requirements; Step 10: Arrange detectors in the target area and analyze and process the detector data to obtain the dispersion energy map E(ω,c); Step 11: Pick the maximum value of the velocity at the corresponding frequency in the dispersion energy diagram E(ω,c) to obtain the dispersion curve f actual ; Step 12: Send the dispersion curve of the target area to the trained model to obtain the prediction of the underground soil density state classification in the target area.

2. The method according to claim 1, wherein In step 3, the method for calculating a theoretical dispersion curve is as follows: According to the underground geological structure of the exploration area, the physical properties of each layer are determined, including the longitudinal wave velocity Vp, the shear wave velocity Vs, the density ρ and the thickness h, and a layered model is established, where Vp = 1.732*Vs; In each layer, it is assumed that the surface wave propagates horizontally and its displacement is in the form of a plane wave, satisfying the elastic dynamics equation: Where u represents the particle displacement vector, ρ is the density, λ and μ are the Lame constants; For Rayleigh waves, the following dispersion equation is derived by combining the boundary condition of zero stress on the free surface and the interlayer continuity condition of displacement and stress continuity: D(ω,k)=0 Where ω is the angular frequency, k is the wave number, c = ω / k, and c is the phase velocity; For Love waves, when only the horizontal displacement component is considered, the dispersion equation is expressed as: in: (1) Select a suitable angular frequency range ω; (2) According to the formation model parameters, set the initial phase velocity c and calculate the corresponding wave number k = ω / c; (3) Use the numerical iteration method to solve the dispersion equation D(ω,k)=0 to obtain the phase velocity c at a specific frequency; (4) Repeat (1), (2), and (3) to obtain the corresponding relationship between angular frequency ω and phase velocity c, that is, the corresponding theoretical dispersion curve.

3. The method according to claim 1, wherein In step seven, the shared feature layer is constructed as follows: each layer includes a one-dimensional convolution layer; the convolution layer is followed by a batch normalization layer; after batch normalization, it is fed into an activation function; the output after the activation function is fed into a regularization layer; and after the regularization layer, it is fed into a pooling layer.

4. The method according to claim 1, wherein In step seven, the embedding layer and the fully connected classification layer in the model are constructed as follows: each layer of soil enters the embedding layer; the shared feature layer and the embedding layer data of the corresponding layer are spliced; the spliced ​​data is sent to the fully connected layer of the corresponding branch for output to obtain the corresponding soil density state classification.

5. The method according to claim 1, wherein In the step nine, the criteria for meeting the requirements are composed of the loss function Loss and the accuracy rate Accuracy, wherein the training process includes a training set and a validation set; the loss function sizes of the training set and the validation set are Loss1 and Loss2; the accuracy rates are Acc1 and Acc2, The respective Loss functions and Acc need to reach certain thresholds ∈1, ∈2, depending on the training set; at the same time, the values ​​of Loss1-Loss2 and Acc1-Acc2 also need to reach certain thresholds ∈3, ∈4, depending on the training situation.

Images