Soil layer dynamics parameter inversion method and device

Abstract

The present disclosure provides a soil layer dynamics parameter inversion method and device, wherein the method comprises: step 1, determining the upper and lower boundaries of the search space corresponding to the shear wave velocity of the shallowest layer of soil, initializing other parameters required for empirical learning and candidate solution set; step 2, updating the candidate solution to generate a new candidate solution set through a random candidate solution and a currently best-performing candidate solution, and retaining the set that performs better between the new candidate solution set and the candidate solution set before updating; step 3, selecting the best-performing candidate solution in the new position candidate solution set, generating a search dimension enhanced candidate solution corresponding to the best-performing candidate solution, retaining a candidate solution that performs better between the currently best-performing candidate solution, and obtaining a search enhanced candidate solution set; step 4, if the search enhanced candidate solution set retained in the current or last iteration is the search dimension enhanced candidate solution, performing secondary enhancement on the search dimension enhanced candidate solution; repeating steps 2 to 4.

Description

Technical Field

[0001] This document relates to the field of computer technology, and in particular to a method and apparatus for inverting soil dynamic parameters. Background Technology

[0002] MASW (Multichannel Analysis of Surface Waves) is a non-destructive testing method for the dynamic parameters of layered media, commonly used for inversion of soil parameters such as shear wave velocity and thickness. The MASW method consists of two steps: the first step is to acquire multichannel records from the testing site to obtain the dispersion curves of the field test; the second step is to invert the dispersion curves to obtain the field soil shear wave velocity model.

[0003] In related technologies, traditional methods often use least squares and other methods for the second step of the MASW inversion, which often leads to drawbacks such as low computational efficiency, susceptibility to local optima, low identification accuracy, and inability to efficiently obtain accurate soil layer parameters. Furthermore, the large search range for the shallowest soil shear wave velocity further hinders the inversion process. Empirical learning methods are intelligent optimization algorithms based on empirical learning that can balance global and local search capabilities, making them suitable for solving complex nonlinear problems similar to soil layer inversion. However, their local search capabilities need improvement.

[0004] Based on the above analysis of the development status of this technology field, the existing technology experience learning methods lack a solution for further optimization based on strengthening the search in the original dimensions. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for inverting soil dynamic parameters, aiming to solve the above-mentioned problems in the prior art.

[0006] According to a first aspect of the present disclosure, a method for inverting soil dynamic parameters is provided, comprising:

[0007] Step 1: Obtain the dispersion curve from the field test. Use the non-dispersion segment of the high-frequency part of the dispersion curve to narrow down and initialize the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil. Initialize other parameters except for the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil. Initialize the candidate solution set in the population.

[0008] Step 2: Select a certain number of candidate solutions from the candidate solution set as random candidate solutions. Update all candidate solutions using the random candidate solutions and the currently best performing candidate solution to generate a new candidate solution set. Retain the set of better performing candidate solutions from the new candidate solution set and the previous candidate solution set as the new position candidate solution set.

[0009] Step 3: Select the best performing candidate solution from the new position candidate solution set, generate a search dimension enhancement candidate solution corresponding to the best performing candidate solution, and retain the better candidate solution between the search dimension enhancement candidate solution and the current best performing candidate solution from the new position candidate solution set to obtain the search enhancement candidate solution set.

[0010] Step 4: If the enhanced candidate solution set in the current iteration or the previous iteration contains a enhanced candidate solution for the search dimension, perform secondary enhancement on the enhanced candidate solution for the search dimension to obtain a secondary enhanced candidate solution set, and use the secondary enhanced candidate solution set as the candidate solution set for the next iteration; otherwise, use the enhanced candidate solution set as the candidate solution set for the next iteration; repeat steps 2 to 4 until the maximum number of iterations set in the initialization is met. Use a custom objective function as the optimization objective in the iteration. After the iteration, select the best performing candidate solution in the final candidate solution set as the inversion value of each decision variable.

[0011] According to a second aspect of the present disclosure, a soil dynamics parameter inversion device is provided, comprising:

[0012] The initialization module is used to obtain the dispersion curve of the field test, reduce the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil by the non-dispersion segment of the high-frequency part of the dispersion curve and initialize it, initialize other parameters except for the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil, and initialize the candidate solution set in the population.

[0013] The multi-mode new position generation module is used to select a certain number of candidate solutions from the candidate solution set as random candidate solutions. By using the random candidate solutions and the currently best performing candidate solution, all candidate solutions are updated to generate a new candidate solution set. The set of better performing candidate solutions in the new candidate solution set and the previous candidate solution set are retained as the new position candidate solution set.

[0014] The dimension search enhancement module is used to select the best performing candidate solution from the new position candidate solution set, generate a search dimension enhancement candidate solution corresponding to the best performing candidate solution, and retain the better candidate solution between the search dimension enhancement candidate solution and the current best performing candidate solution in the new position candidate solution set to obtain the search enhancement candidate solution set.

[0015] The dimensional secondary search enhancement module is used to perform secondary enhancement on the dimensional enhancement candidate solutions when the candidate solution set is determined to contain dimensional enhancement candidate solutions in the current or previous iteration. This results in a secondary enhancement candidate solution set, which is then used as the candidate solution set for the next iteration. Otherwise, the dimensional enhancement candidate solution set is used as the candidate solution set for the next iteration. The multi-mode new position generation module, the dimensional search enhancement module, and the dimensional secondary search enhancement module are repeatedly executed until the maximum number of iterations set in the initialization is met. A custom objective function is used as the optimization objective during the iteration. After the iteration, the best-performing candidate solution in the final candidate solution set is selected as the inversion value of each decision variable.

[0016] The technical solution provided by the embodiments of the present invention has the following beneficial effects: by narrowing the upper and lower bounds of the search space corresponding to the shear wave velocity of the shallowest soil through the non-dispersion segment of the high-frequency part of the dispersion curve, the inversion efficiency is improved; by adding dimensional search for secondary optimization on the basis of the original empirical learning method, the local search capability of the empirical learning method is enhanced, and the candidate solution can approach the optimal solution more quickly.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the soil dynamic parameter inversion method according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of the soil dynamic parameter inversion process according to an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the identification result of a specific embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of the identification result of a specific embodiment two of the present invention;

[0023] Figure 5 This is a schematic diagram of the identification result of a specific embodiment three of the present invention;

[0024] Figure 6This is a schematic diagram of the soil dynamic parameter inversion device according to an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0026] Method Implementation Examples

[0027] According to an embodiment of the present invention, a method for inverting soil dynamic parameters is provided. Figure 1 This is a flowchart of the soil dynamic parameter inversion method according to an embodiment of the present invention, as follows: Figure 1 As shown, the soil dynamic parameter inversion method according to an embodiment of the present invention specifically includes:

[0028] In step S110, the dispersion curve from the field test is obtained. The search space upper and lower bounds of the shallowest soil shear wave velocity are narrowed and initialized by using the non-dispersion segment of the high-frequency part of the dispersion curve. Other parameters besides the upper and lower bounds of the search space for the shallowest soil shear wave velocity are initialized, and the candidate solution set in the population is initialized. Specifically, this includes:

[0029] Formula 1 represents the set of inversion parameters. Formula 2 is used to determine the estimated value V' of the shear wave velocity in the shallowest soil layer. R1 Using formulas 3 and 4 according to V' R1 Narrowing the search space for the shear wave velocity in the shallowest soil layer:

[0030] {p i}={h1 h2 … h n-1 V R1 V R2 … V Rn D1 D2 … D n} Formula 1;

[0031]

[0032]

[0033]

[0034] In the formula, {p i} represents the vector of inversion parameters, h1h2…h n-1V represents the thickness of the soil layers from the shallowest to the deepest, respectively. R1 V R2 …V Rn The numbers D1, D2, ..., D represent the soil shear wave velocities from the shallowest to the deepest layers, respectively. n V represents the soil damping from the shallowest to the deepest layer, respectively. s1 V' represents the shallowest soil phase velocity obtained through the non-dispersion segment of the high-frequency portion of the dispersion curve, σ represents the soil density, and V' represents the density of the soil layer. R1 This represents the estimated shear wave velocity in the shallowest soil layer. This represents the lower bound of the search space corresponding to the shear wave velocity of the shallowest soil layer. This represents the upper bound of the search space corresponding to the shear wave velocity of the shallowest soil layer.

[0035] The parameters for initializing the search space upper and lower bounds, excluding the shear wave velocity of the shallowest soil layer, include: initializing the maximum number of iterations (Itermax), the population size (Np), the number of decision variables (n), and the upper and lower bounds of the search space for each decision variable other than the shear wave velocity of the shallowest soil layer (Xmax and Xmin). The candidate solution set in the population is initialized using Equation 5:

[0036] X p,q =X min +rand(X max -X min ), p=1,2,…,Np,q=1,2,…,n Formula 5;

[0037] In the formula, rand is a random number uniformly distributed in the range [0,1], and X p,q This represents the candidate solution set at all positions of the initialized candidate solution set. The initialized candidate solution set is similar to a p×q matrix.

[0038] In step S120, a certain number of candidate solutions are selected from the candidate solution set as random candidate solutions. Using these random candidate solutions and the currently best-performing candidate solution, all candidate solutions are updated to generate a new candidate solution set. The set of candidate solutions that outperforms the previous set is retained and used as the new position candidate solution set. Specifically, this includes:

[0039] Three candidate solutions are randomly selected from the candidate solution set as random candidate solutions, namely X. l,q X m,q X n,q The best candidate solution currently is X. best,qThe algorithm determines the mode to be used for updating candidate solutions. If the dynamically adjusted parameter calculated by the number of iterations is greater than or equal to the random number used for this update, a cautious search mode is used to update all candidate solutions; otherwise, a random search mode is used. Formula 6 indicates that a cautious search mode is used to update all candidate solutions, Formula 7 indicates the expression for the iterative change parameter AF in Formula 6, and Formula 8 indicates that a random search mode is used to update all candidate solutions.

[0040]

[0041]

[0042]

[0043] In the formula, This represents the candidate solution after updating all positions, AF represents the iteration change parameters, and Iter represents the number of iterations. max Let Itermax represent the maximum number of iterations, exp represent the exponential function, and obj represent the performance evaluation function. After updating the candidate solutions for all positions, a new set of candidate solutions for each position is obtained. A dynamically adjusted parameter NF is introduced to adjust the use of the two modes in each iteration. When NF is greater than or equal to rand, the cautious search mode is used to update X. p,q When NF is less than rand, update X using a random search pattern. p,q Calculate the NF value using Equation 9, and update all candidate solutions using Equation 10:

[0044]

[0045]

[0046] In step S130, the best-performing candidate solution in the new position candidate solution set is selected, and a search dimension enhancement candidate solution corresponding to the best-performing candidate solution is generated. Then, in the new position candidate solution set, the candidate solution that performs better than the current best-performing candidate solution is retained, resulting in a search enhancement candidate solution set. Specifically, this includes:

[0047] To generate candidate solutions for each dimension of the best-performing candidate solution, candidate solutions for the corresponding search dimensions are generated by considering the step size of the current iteration. Equation 11 is used to generate candidate solutions for the strengthened search dimensions, and Equation 12 represents the expression for the iteration step size in Equation 11. Equation 13 is used to select the better candidate solution between the strengthened search dimension candidate solution and the current best-performing candidate solution.

[0048]

[0049]

[0050]

[0051] In the formula, Let X represent the step size of the iterth iteration. best,q This represents the best candidate solution currently being performed. This indicates that the generated search dimension strengthens the candidate solutions, X worst,q This represents the worst candidate solution currently, where X... p,q This indicates that the retained search dimensions strengthen the candidate solution and make it the better candidate solution among the current best performing candidate solutions.

[0052] In step S140, if the candidate solution set for strengthening the search dimension is retained in the current iteration or the previous iteration, the candidate solution for strengthening the search dimension is strengthened a second time to obtain a secondary strengthened candidate solution set, which is then used as the candidate solution set for the next iteration; otherwise, the candidate solution set for strengthening the search dimension is used as the candidate solution set for the next iteration; steps S120 to S40 are repeated until the maximum number of iterations set in the initialization is met. A custom objective function is used as the optimization objective during the iterations. After the iterations, the best-performing candidate solution in the final candidate solution set is selected as the inversion value of each decision variable. Specifically, this includes:

[0053] Depending on whether the enhanced candidate solution for preserving the search dimension occurred in the current iteration or the previous iteration, different calculation methods are used to obtain the attenuation coefficient. The enhanced candidate solution for the search dimension is then further enhanced by considering the step size of the current iteration and the attenuation coefficient. Equation 14 is used for this secondary enhancement, and Equation 15 expresses the expression for the attenuation coefficient in Equation 14:

[0054]

[0055]

[0056] In the formula, Re represents the step size of the iterth iteration. q Indicates the attenuation coefficient. This indicates that the retained search dimensions strengthen the candidate solutions, i.e., X generated in step S130. p,q , This represents a candidate solution that is further enhanced by strengthening the candidate solution for the search dimension. The set of candidate solutions for secondary enhancement is obtained by updating the candidate solutions for strengthening the search dimension.

[0057] Using a custom objective function as the optimization objective in iterations specifically includes using objective functions based on dispersion curves, dissipation curves, and Arias intensity curves during the iteration process. Equation 16 represents the expression of the custom objective function:

[0058]

[0059] In the formula, f obj (p i ) represents the objective function, N c C represents an equidistant frequency range of 8-60Hz. R A R I zz These are represented as dispersion curves, dissipation curves, and Arias intensity curves, respectively; the superscript T indicates the theoretical value, and the superscript E indicates the experimental value; α, β, and γ represent the proportion of each objective function in the inversion process, with α ranging from 0.8 to 1, β from 0.3 to 0.5, and γ from 0.5 to 0.8; p i The vector {p} represents the inversion parameters i The inversion parameters in}; k represents the k-th detector, r k This represents the distance between the k-th detector and the vibration source.

[0060] Figure 2 This is a schematic diagram of the soil dynamic parameter inversion process according to an embodiment of the present invention, as shown below. Figure 2 As shown, this embodiment demonstrates the complete process of inverting soil dynamic parameters using an improved empirical learning method.

[0061] In the first specific embodiment, a homogeneous soil layer with increasing shear wave velocity was used. The maximum number of iterations (Itermax) was set to 300, and the population size (Np) was set to 40. The basic information of the homogeneous soil layer with increasing shear wave velocity is shown in Table 1, and the inversion results are shown in Table 2.

[0062] Table 1. Basic Information of Example 1

[0063] First layer 3 150 300 1900 Second floor 4 250 500 1900 Third layer ∞ 350 700 1900

[0064] Table 2. Inversion Results of Example 1

[0065]

[0066] Figure 3 This is a schematic diagram of the identification result of a specific embodiment of the present invention, as shown in the figure. Figure 3 As shown, the method used in this embodiment of the invention is superior to the traditional empirical learning method in terms of recognition speed and accuracy, and the inversion parameters of the recognition are closer.

[0067] In the second specific embodiment, a homogeneous soil layer containing weak interlayers was used. The maximum number of iterations (Itermax) was set to 300, and the population size (Np) was set to 40. The basic information of the homogeneous soil layer containing weak interlayers is shown in Table 3, and the inversion results are shown in Table 4.

[0068] Table 3. Basic Information of Example 2

[0069] First layer 3 250 500 1900 Second floor 4 150 300 1900 Third layer ∞ 350 700 1900

[0070] Table 4. Inversion Results of Example 2

[0071]

[0072] Figure 4 This is a schematic diagram of the identification result of a specific embodiment two of the present invention, as shown below. Figure 4 As shown, the dispersion curve of the method used in this embodiment of the invention is closer to the theoretical value than that of the traditional empirical learning method, and the identified inversion parameters are also closer.

[0073] In specific embodiment three, a homogeneous soil layer with hard interlayers was used, the maximum number of iterations Itermax was set to 300, the population size Np was set to 40, the basic information of the homogeneous soil layer with weak interlayers is shown in Table 5, and the inversion results are shown in Table 6:

[0074] Table 5. Basic Information of Example 3

[0075]

[0076]

[0077] Table 6. Inversion Results of Example 3

[0078]

[0079] Figure 5 This is a schematic diagram of the identification result of a specific embodiment three of the present invention, as shown below. Figure 5 As shown, the method used in this embodiment of the invention is superior to the traditional empirical learning method in terms of recognition speed and accuracy, and the inversion parameters of the recognition are closer.

[0080] In summary, to address the existing problems, this invention proposes a method for inverting soil dynamic parameters. By narrowing the upper and lower bounds of the search space corresponding to the shear wave velocity of the shallowest soil through the non-dispersion segment of the high-frequency part of the dispersion curve, the inversion efficiency is improved. Based on the original empirical learning method, a secondary optimization of dimensional search is added. When the dimensional search finds a better optimization direction, a secondary search is attempted in that direction. If the secondary search yields a better candidate solution, the search is repeated in that direction during the second iteration; otherwise, the search radius is reduced before searching in that direction again. This enhances the local search capability of the empirical learning method, enabling candidate solutions to approach the optimal solution more quickly and efficiently.

[0081] Device Examples

[0082] According to an embodiment of the present invention, a soil dynamic parameter inversion device is provided. Figure 6This is a schematic diagram of the soil dynamic parameter inversion device according to an embodiment of the present invention, as shown below. Figure 6 As shown, the soil dynamic parameter inversion device according to an embodiment of the present invention specifically includes:

[0083] The initialization module 60 is used to obtain the dispersion curve of the field test, reduce the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil by the non-dispersion segment of the high-frequency part of the dispersion curve and initialize it, initialize other parameters except for the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil, and initialize the candidate solution set in the population.

[0084] The other parameters besides the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil include: the maximum number of iterations, the number of population sizes, the number of decision variables, and the upper and lower bounds of the search space for each decision variable other than the shear wave velocity of the shallowest soil.

[0085] The multi-mode new position generation module 62 is used to select a certain number of candidate solutions from the candidate solution set as random candidate solutions. Using these random candidate solutions and the currently best-performing candidate solution, it updates all candidate solutions to generate a new candidate solution set. The set of candidate solutions that outperforms the previous set is retained as the new position candidate solution set. Specifically, the multi-mode new position generation module 62 is used for:

[0086] Three candidate solutions are randomly selected from the candidate solution set as random candidate solutions, and the mode used for updating candidate solutions in this update is determined. That is, the dynamic adjustment parameter is calculated by the number of iterations. If the dynamic adjustment parameter is greater than or equal to the random number used in this update, all candidate solutions are updated using the cautious search mode; otherwise, all candidate solutions are updated using the random search mode.

[0087] The dimensional search enhancement module 64 is used to select the best-performing candidate solution from the new position candidate solution set, generate a search dimension-enhanced candidate solution corresponding to the best-performing candidate solution, and retain the better candidate solution between the search dimension-enhanced candidate solution and the current best-performing candidate solution from the new position candidate solution set, thus obtaining the search enhancement candidate solution set. Specifically, the dimensional search enhancement module 64 is used for:

[0088] To generate candidate solutions for each dimension that represents the best candidate solution, candidate solutions are generated at corresponding positions. The candidate solutions are strengthened by generating corresponding search dimensions that take into account the step size of the current iteration.

[0089] The dimensional secondary search enhancement module 66 is used to perform secondary enhancement on the dimensional enhancement candidate solutions when the candidate solution set is determined to contain dimensional enhancement candidate solutions in the current or previous iteration, resulting in a secondary enhancement candidate solution set, which is then used as the candidate solution set for the next iteration. Otherwise, the dimensional enhancement candidate solution set is used as the candidate solution set for the next iteration. The multi-mode new position generation module, the dimensional search enhancement module, and the dimensional secondary search enhancement module are repeatedly executed until the maximum number of iterations set in the initialization is met. A custom objective function is used as the optimization objective during iteration. After the iteration, the best-performing candidate solution in the final candidate solution set is selected as the inversion value for each decision variable. Specifically, the dimensional secondary search enhancement module 66 is used for:

[0090] Depending on whether the enhanced candidate solution for preserving the search dimension occurred in the current iteration or the previous iteration, different calculation methods are used to obtain the attenuation coefficient. The enhanced candidate solution for the search dimension is then further enhanced by considering the step size of the current iteration and the attenuation coefficient.

[0091] The custom objective function used in the iteration is determined based on the dispersion curve, dissipation curve, and Arias intensity curve.

[0092] In summary, to address the existing problems, this invention proposes a soil dynamics parameter inversion device. By narrowing the upper and lower bounds of the search space corresponding to the shear wave velocity of the shallowest soil through the non-dispersion segment of the high-frequency part of the dispersion curve, the inversion efficiency is improved. Based on the original empirical learning method, a secondary optimization through dimensional search is added. When the dimensional search finds a better optimization direction, a secondary search is attempted in that direction. If the secondary search yields a better candidate solution, the search is repeated in that direction during the second iteration; otherwise, the search radius is reduced before searching in that direction again. This enhances the local search capability of the empirical learning method, enabling candidate solutions to approach the optimal solution more quickly and efficiently.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for inverting soil dynamic parameters, characterized in that, include: Step 1: Obtain the dispersion curve from the field test. Narrow the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil by using the non-dispersion segment of the high-frequency part of the dispersion curve and initialize it. Initialize other parameters except for the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil. Initialize the candidate solution set in the population. Step 2: Select a certain number of candidate solutions from the candidate solution set as random candidate solutions. Update all candidate solutions using the random candidate solutions and the currently best performing candidate solution to generate a new candidate solution set. Retain the set of candidate solutions that outperforms the candidate solution set before the update and use it as the new position candidate solution set. Step 3: Select the best performing candidate solution from the new position candidate solution set, generate a search dimension enhancement candidate solution corresponding to the best performing candidate solution, and retain the better candidate solution between the search dimension enhancement candidate solution and the current best performing candidate solution in the new position candidate solution set to obtain the search enhancement candidate solution set. Step 4: If the search-enhanced candidate solution set retained in the current iteration or the previous iteration is the search-dimension-enhanced candidate solution, perform secondary enhancement on the search-dimension-enhanced candidate solution to obtain a secondary enhanced candidate solution set, and use the secondary enhanced candidate solution set as the candidate solution set for the next iteration; otherwise, use the search-enhanced candidate solution set as the candidate solution set for the next iteration; repeat steps 2 to 4 until the maximum number of iterations set in the initialization is met. In the iteration, a custom objective function is used as the optimization objective. After the iteration, the best-performing candidate solution in the final candidate solution set is selected as the inversion value of each decision variable.

2. The method according to claim 1, characterized in that, The parameters for initializing the search space upper and lower bounds other than the shear wave velocity of the shallowest soil layer specifically include: the maximum number of iterations, the number of population sizes, the number of decision variables, and the upper and lower bounds of the search space corresponding to each decision variable other than the shear wave velocity of the shallowest soil layer.

3. The method according to claim 1, characterized in that, The step of selecting a certain number of candidate solutions from the candidate solution set as random candidate solutions, and updating all candidate solutions to generate a new candidate solution set using the random candidate solutions and the currently best-performing candidate solution, specifically includes: Three candidate solutions are randomly selected from the candidate solution set as random candidate solutions, and the mode used for updating the candidate solutions is determined. If the dynamic adjustment parameter calculated by the number of iterations is greater than or equal to the random number used for this update, all candidate solutions are updated using the cautious search mode; otherwise, all candidate solutions are updated using the random search mode.

4. The method according to claim 1, characterized in that, The enhanced candidate solutions corresponding to the best performing candidate solutions specifically include: To generate candidate solutions for each dimension that represents the best candidate solution, candidate solutions are generated at corresponding positions. The candidate solutions are strengthened by generating corresponding search dimensions that take into account the step size of the current iteration.

5. The method according to claim 1, characterized in that, If the search enhancement candidate solution set retained in the current iteration or the previous iteration is the search dimension enhancement candidate solution, then the search dimension enhancement candidate solution is further enhanced to obtain the secondary enhancement candidate solution set, specifically including: Depending on whether the enhanced candidate solution for the search dimension occurred in the current iteration or the previous iteration, different calculation methods are used to obtain the attenuation coefficient. The enhanced candidate solution for the search dimension is then further enhanced by considering the step size of the current iteration and the attenuation coefficient.

6. The method according to claim 1, characterized in that, The use of a custom objective function as the optimization objective in the iteration specifically includes using an objective function based on dispersion curves, dissipation curves, and Arias intensity curves during the iteration process.

7. A soil dynamic parameter inversion device, characterized in that, include: The initialization module is used to obtain the dispersion curve of the field test, reduce the upper and lower bounds of the search space of the shallowest soil shear wave velocity by the non-dispersion segment of the high-frequency part of the dispersion curve and initialize it, initialize other parameters except the upper and lower bounds of the search space of the shallowest soil shear wave velocity, and initialize the candidate solution set in the population. The multi-mode new position generation module is used to select a certain number of candidate solutions from the candidate solution set as random candidate solutions, update all candidate solutions to generate a new candidate solution set using the random candidate solutions and the currently best performing candidate solution, and retain the set of better performing candidates solution sets in the new candidate solution set and the previous candidate solution set as the new position candidate solution set. The dimension search enhancement module is used to select the best performing candidate solution from the new position candidate solution set, generate a search dimension enhancement candidate solution corresponding to the best performing candidate solution, and retain the better candidate solution between the search dimension enhancement candidate solution and the current best performing candidate solution in the new position candidate solution set to obtain a search enhancement candidate solution set. The dimensional secondary search enhancement module is used to perform secondary enhancement on the search dimension enhancement candidate solution when it is determined in the current or previous iteration that the search enhancement candidate solution set contains the search dimension enhancement candidate solution. This results in a secondary enhancement candidate solution set, which is then used as the candidate solution set for the next iteration. Otherwise, the search enhancement candidate solution set is used as the candidate solution set for the next iteration. The multi-mode new position generation module, the dimensional search enhancement module, and the dimensional secondary search enhancement module are repeatedly executed until the maximum number of iterations set in the initialization is met. A custom objective function is used as the optimization objective during the iteration. After the iteration, the best-performing candidate solution in the final candidate solution set is selected as the inversion value of each decision variable.

8. The apparatus according to claim 7, characterized in that, The parameters other than the upper and lower bounds of the search space for the shear wave velocity of the shallowest soil include: the maximum number of iterations, the number of population sizes, the number of decision variables, and the upper and lower bounds of the search space for each decision variable other than the shear wave velocity of the shallowest soil.

9. The apparatus according to claim 7, characterized in that, The multi-mode new position generation module is specifically used to: randomly select three candidate solutions from the candidate solution set as random candidate solutions, and determine the mode used for updating the candidate solutions this time, that is, calculate the dynamic adjustment parameter by the number of iterations. If the dynamic adjustment parameter is greater than or equal to the random number used for this update, use the cautious search mode to update all candidate solutions; otherwise, use the random search mode to update all candidate solutions. The dimension search enhancement module is specifically used to: generate candidate solutions for each dimension of the best performing candidate solution, and generate corresponding search dimension enhancement candidate solutions by considering the step size of the current iteration; The dimension-enhancing module is specifically used to: obtain the attenuation coefficient by using different calculation methods based on whether the candidate solution for enhancing the search dimension occurred in the current iteration or the previous iteration, and perform secondary enhancement on the candidate solution for enhancing the search dimension by considering the step size of the current iteration and the attenuation coefficient.

10. The apparatus according to claim 7, characterized in that, The custom objective function is determined based on the dispersion curve, dissipation curve, and Arias intensity curve.

Images