A geotechnical body physical mechanics parameter coupling inversion survey method
By collaboratively acquiring multi-source data and synchronizing it in time and space, a coupled model was established and an improved adaptive particle swarm optimization algorithm was adopted. This solved the systematic error and local optimum problem of parameter coupling relationship in the investigation of physical and mechanical parameters of soil and rock, and achieved high-precision inversion of physical and mechanical parameters of soil and rock.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QINGHAI ELECTRIC POWER DESIGN INST
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-24
AI Technical Summary
Existing methods for investigating the physical and mechanical parameters of soil and rock masses ignore the inherent coupling relationship between parameters, resulting in systematic errors in the calculation results, low data utilization, and the inversion process is prone to getting trapped in local optima, and cannot meet the needs of fine investigation.
By adopting multi-source data collaborative acquisition and spatiotemporal synchronization, a coupled model is established through outlier removal, normalization, and adaptive weighted fusion. An improved adaptive particle swarm optimization algorithm is then used to perform multi-dimensional constraint inversion solution, quantify parameter coupling relationships, and achieve high-precision inversion.
It improves the accuracy and data utilization of inversion of physical and mechanical parameters of soil and rock, avoids parameter outliers, enhances the reliability of inversion results and the accuracy of engineering verification, and solves the problem of local optima in high-dimensional parameter space.
Smart Images
Figure CN122451289A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of geotechnical engineering investigation technology, specifically to a method for coupled inversion investigation of physical and mechanical parameters of soil and rock. Background Technology
[0002] The physical and mechanical parameters of soil and rock masses are the core basis for engineering investigation and design, stability evaluation, and disaster early warning; the accuracy of their values directly determines the safety and economy of the project. Current methods for investigating soil and rock parameters mainly suffer from the following shortcomings: First, traditional methods often use an independent inversion mode, treating each physical and mechanical parameter as an independent variable and ignoring the inherent coupling relationship between the parameters, which leads to systematic errors in the calculation results.
[0003] Second, multi-source data such as drilling, in-situ testing, laboratory experiments, geophysical exploration, and dynamic monitoring often use a single data source for inversion or simple data splicing, without achieving effective spatiotemporal fusion. Data is used in isolation or simply spliced, resulting in low utilization and poor consistency.
[0004] Third, the inversion process is constrained only by data fitting, lacking multi-dimensional limitations such as geological laws, engineering experience, and physical meaning, which can easily lead to abnormal results that do not conform to reality (such as negative internal friction angle and permeability coefficient far exceeding the range of lithological theory). Fourth, traditional inversion methods often employ gradient descent and standard particle swarm optimization, which are prone to getting trapped in local optima in high-dimensional parameter spaces and are sensitive to initial parameters. They also have low convergence accuracy for complex rock and soil bodies and cannot meet the needs of fine exploration.
[0005] Therefore, there is an urgent need for a coupled inversion method for physical and mechanical parameters of soil and rock to solve the problems mentioned above. Summary of the Invention
[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for coupled inversion exploration of physical and mechanical parameters of soil and rock, so as to realize the collaborative utilization of multi-source data, quantification of parameter coupling relationships, multi-dimensional constraint control and high-precision inversion solution.
[0007] To achieve the above objectives, this application adopts the following technical solution: A method for coupled inversion investigation of physical and mechanical parameters of soil and rock masses includes the following steps: Step 1: Collaborative acquisition of multi-source exploration data, determination of data types and spatiotemporal synchronous acquisition, and establishment of a five-dimensional data ledger; Step 2: Multi-source data preprocessing and spatiotemporal fusion, obtaining a unified data source through outlier removal, normalization, and adaptive weighted fusion; Step 3: Model the coupling relationship between physical and mechanical parameters of soil and rock, select coupling parameter pairs, quantify coupling coefficients and construct coupling models; Step 4: Construct the coupled inversion objective function, integrating data fitting error, coupling constraint error, and regularization term; Step 5: Set multi-dimensional inversion constraints, including geological conditions, engineering experience, physical significance, and dynamic monitoring constraints; Step 6: Design of an improved adaptive particle swarm optimization algorithm, employing adaptive inertia weights, dynamic learning factors, and crowding control; Step 7: Coupled inversion iterative solution and parameter update, iteratively optimize parameters and dynamically update the coupled model; Step 8: Verify the reliability of the inversion results and generate the survey report. Output the report through residual, cross and engineering verification.
[0008] The physical and mechanical parameters of the soil and rock mass described in this application include elastic modulus E, Poisson's ratio μ, cohesion c, internal friction angle φ, permeability coefficient k, and porosity g.
[0009] Furthermore, the data types mentioned in step 1 include geological drilling data, in-situ test data, laboratory test data, geophysical data, and dynamic monitoring data; a GNSS positioning system is used to synchronize the spatial coordinates of all data, constructing a five-dimensional data ledger containing data type, spatial coordinates, acquisition time, testing instruments, and operators, as detailed below: (1) Data type determination: Based on the rock and soil type of the survey area and the engineering requirements, determine the data types to be collected, including: Geological drilling data: Physical property parameters of core samples, including natural density. and moisture content Sampling interval ≤ 1.0m; In-situ test data: Standard Penetration Test (SPT) blow count (N), Static Penetration Test (CPT) cone tip resistance. vane shear strength The test depth interval is ≤0.5m; Indoor test data: Elastic modulus E and Poisson's ratio from triaxial compression tests Cohesion in direct shear test Internal friction angle φ, with ≥3 lithological samples per group; Geophysical data: P-wave velocity from seismic wave velocity testing transverse wave velocity Apparent resistivity of resistivity test The spacing between survey lines is ≤5.0m; Dynamic monitoring data: time series data of slope displacement and foundation pit settlement; (2) Spatiotemporal synchronous acquisition: The spatial coordinates of all data are synchronized using a GNSS positioning system; environmental parameters at the time of acquisition are marked on the dynamic monitoring data; (3) Data ledger establishment: Construct a five-dimensional ledger that includes data type, spatial coordinates, collection time, testing instruments and operators to ensure data traceability.
[0010] Furthermore, step 2 specifically includes: (1) Outlier removal: Outliers are identified using the Z-score method, and the calculation formula is as follows:
[0011] in, For the first One data value, The mean of the data. The standard deviation is denoted as ; when If a value is identified as an outlier, it is replaced by linear interpolation of adjacent data. If the remaining amount after removing a single type of data is less than 50%, the original data is retained and marked for review. (2) Data normalization: Normalize data of different dimensions to the interval [0,1], using the following formula:
[0012] in, , Divided into minimum and maximum data values, The data is after normalization; (3) Spatiotemporal fusion: An adaptive weighted fusion model is used to fuse multi-source data from the same spatial location and time period. The formula is:
[0013] in, The merged data values; The number of data sources, 1 ≤ ≤5; For the first Weight coefficients of data sources Based on data credibility Spatiotemporal matching degree Sure, ; For data reliability, the values are 0.9-1.0 for indoor tests, 0.7-0.9 for in-situ tests, and 0.6-0.8 for geophysical exploration. For spatiotemporal matching degree, =1 / (1+d k ), d k The Euclidean distance between the coordinates of the k-th data source and the target point is calculated. For the first Data source in spatial coordinates ,time The data value.
[0014] Furthermore, step 3 specifically includes: (1) Coupling parameter selection: Based on engineering mechanics theory and laboratory test data, core coupling parameter pairs including elastic modulus E and Poisson's ratio were selected. Elastic parameter coupling, cohesion Coupled with the strength parameter of the internal friction angle φ and the permeability coefficient With porosity Seepage-physical parameter coupling: (2) Coupling relationship quantification: Pearson correlation analysis is used to establish the coupling correlation matrix, the formula is:
[0015] in, For parameters and The coupling coefficient has a value range of [-1, 1]. A value ≥0.6 indicates strong coupling; This refers to the number of samples in the indoor test. ≥30; For the first The first sample Parameter values, For the first The mean of each parameter; the coupling correlation matrix , For parameter dimensions, ≥3; (3) Coupled model construction: A parameter coupling model is established based on the coupling coefficients, as follows: The elastic parameter coupling model is: E=C Eμ ·μ+a; The strength parameter coupling model is: c = Cc φ ·φ+b; seepage The physical parameter coupling model is: k=C kg ·g+e; Where a, b, and c are intercept terms, obtained by linear regression fitting. Further, the coupled inversion objective function constructed in step 4 is:
[0016] in, The parameter vector to be inverted; , , These are the weighting coefficients. Value range: ∈[0.6,0.8], ∈[0.1,0.3], ∈[0.05,0.15]; The data fitting error is calculated using the following formula: , in, To integrate data volume, For measured fusion data, For parameters The corresponding calculated data; The coupling constraint error is calculated using the following formula: , For parameters Calculated coupling coefficients; For regularization terms: This is to prevent parameter overfitting.
[0017] Furthermore, step 5 specifically includes: (1) Geological constraints: The range of parameter values is determined based on the lithology of the exploration area, such as cohesive soil. ∈[5,50]kPa, φ∈[10°,30°]; sandstone E∈[10,50]GPa; (2) Engineering experience constraints: Refer to the parameter range of similar projects in the same area, such as the silty clay of highway slopes. ∈[10 -8 10 -6 ]m / s; (3) Physical constraints: The parameter values must satisfy physical laws, such as ∈[0.15,0.45], ≥0°, >0.
[0018] (4) Dynamic monitoring constraints: The time-series deformation data satisfies , To improve monitoring accuracy.
[0019] Furthermore, step 6 specifically includes: (1) Algorithm improvement strategy: Adaptive inertia weights: , For the number of iterations, This represents the maximum number of iterations. Dynamic learning factor: , ; Crowding control: Calculate the Euclidean distance between particles; when the distance is less than a threshold... , When the value is 0.01, the particle position is randomly reset; (2) Algorithm execution flow: Initialization: Particle swarm size N∈[50,100], parameter search space is determined by the constraints in step 5; Fitness calculation: based on the objective function This is the fitness value; Particle Update: Velocity ,Location , , A random number in the range [0,1]. For the individual's optimal, It is the global optimum; Convergence criterion: 5 consecutive generations Change < 1×10 -4 or reaching the maximum number of iterations. ,100≤ ≤200.
[0020] Furthermore, step 7 specifically includes: (1) Initial parameter setting: The mean value of the indoor test was used as the initial parameter vector. ; (2) Iterative solution: First iteration: Substitute into the coupling model and calculate The objective function is solved using an algorithm to obtain... ; Iterative update: Every 5 iterations, based on the current parameters Recalculate coupling coefficients Update the coupling model; Constraint verification: For each iteration... Constraint checks are performed, and if the constraints are not met, the fitness value is corrected using the penalty function method. (3) Parameter convergence: When the convergence criterion is met, the optimal parameter vector is output. .
[0021] Furthermore, step 8 specifically includes: (1) Reliability verification Reliability verification includes residual analysis, leave-one-out cross-validation, and FLAC. 3D. Numerical simulation engineering verification; three indicators must be met simultaneously: residual mean ≤ 0.05 and standard deviation ≤ 0.1, cross-validation accuracy ≥ 90%, and relative error between calculated deformation value and monitored value ≤ 10%. Only if all bidding parties meet these requirements can a survey report be generated, as detailed below: Residual analysis: calculation The residual mean is ≤0.05 and the standard deviation is ≤0.1. Cross-validation: Using the leave-one-out method, 10% of the data is removed each time during the inversion process, and the inversion accuracy is ≥90%. Engineering verification: Substitute into the numerical simulation software FLAC 3 D, The relative error between the calculated deformation value and the monitored value is ≤10%; (2) Survey report generation: includes parameter inversion results Including confidence intervals, coupling relationship graphs, data fusion quality assessment, reliability verification conclusions, and engineering application recommendations.
[0022] Compared with the prior art, the present invention has the following beneficial effects: This invention addresses the systematic error problem of traditional independent inversion by quantifying the coupling relationship between parameters, significantly improving inversion accuracy compared to existing methods and reducing engineering verification errors. The adaptive weighted fusion model considers data reliability and spatiotemporal matching, significantly improving data utilization compared to single data sources, with a mean residual of ≤0.05. By integrating geological, empirical, physical, and dynamic monitoring constraints, outlier parameters are avoided, ensuring the inversion results conform to actual geological patterns. The adaptive particle swarm optimization algorithm achieves faster convergence than standard algorithms, resolving the local optima problem in high-dimensional parameter spaces. Attached Figure Description
[0023] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0024] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0025] Example 1 A method for coupled inversion investigation of physical and mechanical parameters of soil and rock masses includes the following steps: Step 1: Collaborative acquisition of multi-source exploration data, determination of data types and simultaneous spatiotemporal acquisition, and establishment of a five-dimensional data ledger, as detailed below: (1) Data type determination: Based on the rock and soil type of the survey area and the engineering requirements, determine the data type to be collected. Specific methods include: Geological drilling data: Physical property parameters of core samples, including natural density. and moisture content Sampling interval ≤ 1.0m; In-situ test data: Standard Penetration Test (SPT) blow count (N), Static Penetration Test (CPT) cone tip resistance. vane shear strength The test depth interval is ≤0.5m; Indoor test data: Elastic modulus E and Poisson's ratio from triaxial compression tests Cohesion in direct shear test Internal friction angle φ, with ≥3 lithological samples per group; Geophysical data: P-wave velocity from seismic wave velocity testing transverse wave velocity Apparent resistivity of resistivity test The spacing between survey lines is ≤5.0m; Dynamic monitoring data: time series data of slope displacement and foundation pit settlement; (2) Spatiotemporal synchronous acquisition: The spatial coordinates of all data are synchronized using a GNSS positioning system; environmental parameters at the time of acquisition are marked on the dynamic monitoring data; (3) Data ledger establishment: Construct a five-dimensional ledger that includes data type, spatial coordinates, collection time, testing instruments and operators to ensure data traceability.
[0026] Step 2: Multi-source data preprocessing and spatiotemporal fusion. A unified data source is obtained through outlier removal, normalization, and adaptive weighted fusion. The specific methods are as follows: (1) Outlier removal: Outliers are identified using the Z-score method, and the calculation formula is as follows:
[0027] in, For the first One data value, The mean of the data. The standard deviation is denoted as ; when If a value is identified as an outlier, it is replaced using linear interpolation of adjacent data. (2) Data normalization: Normalize data of different dimensions to the interval [0,1], using the following formula:
[0028] in, , Divided into minimum and maximum data values, The data is after normalization; (3) Spatiotemporal fusion: An adaptive weighted fusion model is used to fuse multi-source data from the same spatial location and time period. The formula is:
[0029] in, The merged data values; The number of data sources, 1 ≤ ≤5; For the first Weight coefficients of data sources Based on data credibility Spatiotemporal matching degree Sure, ; For data reliability, the values are 0.9-1.0 for indoor tests, 0.7-0.9 for in-situ tests, and 0.6-0.8 for geophysical exploration. For spatiotemporal matching degree, =1 / (1+d k ), d k The Euclidean distance between the coordinates of the k-th data source and the target point is calculated.
[0030] For the first Data source in spatial coordinates ,time The data value.
[0031] Step 3: Model the coupling relationship between physical and mechanical parameters of soil and rock, select coupling parameter pairs, quantify coupling coefficients and construct coupling models; (1) Coupling parameter selection: Based on engineering mechanics theory and laboratory test data, the core coupling parameter pairs selected include: Elastic parameter coupling: elastic modulus E and Poisson's ratio ; Strength parameter coupling: cohesion With the internal friction angle φ; Seepage-physical parameter coupling: permeability coefficient With porosity ; (2) Coupling relationship quantification: Pearson correlation analysis is used to establish the coupling correlation matrix, the formula is:
[0032] in, For parameters and The coupling coefficient has a value range of [-1, 1]. A value ≥0.6 indicates strong coupling; This refers to the number of samples in the indoor test. ≥30; For the first The first sample Parameter values, For the first The mean of each parameter; the coupling correlation matrix , For parameter dimensions, ≥3; (3) Coupled model construction: Based on the coupling coefficient, a parametric coupling model is established, such as the elastic parametric coupling model: E=C Eμ ·μ+a; The intensity parameter coupling model is: c=C cφ ·φ+b; seepage The physical parameter coupling model is: k=C kg ·g+e; where a, b, and c are intercept terms, obtained by linear regression fitting.
[0033] Furthermore, the coupled inversion objective function constructed in step 4 is: , which is obtained by linear regression fitting.
[0034] Step 4: Construct the coupled inversion objective function, integrating data fitting error, coupling constraint error, and regularization term; The constructed coupled inversion objective function is:
[0035] in, The parameter vector to be inverted; , , These are the weighting coefficients. Value range: ∈[0.6,0.8], ∈[0.1,0.3], ∈[0.05,0.15]; Let $\mathbf{a}$ be the data fitting error. The formula for calculating the data fitting error is:
[0036] To integrate data volume, For measured fusion data, For parameters The corresponding calculated data; The coupling constraint error is calculated using the following formula: , For parameters Calculated coupling coefficients; For regularization terms: This is to prevent parameter overfitting.
[0037] Step 5: Set multi-dimensional inversion constraints, including geological conditions, engineering experience, physical meaning, and dynamic monitoring constraints, as detailed below: (1) Geological constraints: The range of parameter values is determined based on the lithology of the exploration area, such as cohesive soil. ∈[5,50]kPa, ∈[10°,30°]; Sandstone E∈[10,50]GPa; (2) Engineering experience constraints: Refer to the parameter range of similar projects in the same area, such as the silty clay of highway slopes. ∈[10 -8 10 -6 ]m / s; (3) Physical constraints: The parameter values must satisfy physical laws, such as ∈[0.15,0.45], ≥0°, >0.
[0038] (4) Dynamic monitoring constraints: The time-series deformation data satisfies , To improve monitoring accuracy.
[0039] Step 6: Design of an improved adaptive particle swarm optimization algorithm, employing adaptive inertia weights, dynamic learning factors, and crowding control, as detailed below: (1) Algorithm improvement strategy: Adaptive inertia weights: , For the number of iterations, This represents the maximum number of iterations. Dynamic learning factor: , ; Crowding control: Calculate the Euclidean distance between particles; when the distance is less than a threshold... , When the value is 0.01, the particle position is randomly reset; (2) Algorithm execution flow: Initialization: Particle swarm size N∈[50,100], parameter search space is determined by the constraints in step 5; Fitness calculation: based on the objective function This is the fitness value; Particle Update: Velocity ,Location , , A random number in the range [0,1]. For the individual's optimal, It is the global optimum; Convergence criterion: 5 consecutive generations Change < 1×10 -4 or reaching the maximum number of iterations. ,100≤ ≤200.
[0040] Step 7: Coupled inversion iterative solution and parameter update, iteratively optimizing parameters and dynamically updating the coupled model, specifically including: (1) Initial parameter setting: The mean value of the indoor test was used as the initial parameter vector. ; (2) Iterative solution: First iteration: Substitute into the coupling model and calculate The objective function is solved using an algorithm to obtain... ; Iterative update: Every 5 iterations, based on the current parameters Recalculate coupling coefficients Update the coupling model; Constraint verification: For each iteration... Constraint checks are performed, and if the constraints are not met, the fitness value is corrected using the penalty function method. (3) Parameter convergence: When the convergence criterion is met, the optimal parameter vector is output. .
[0041] Step 8: Verify the reliability of the inversion results and generate the survey report. Output the report through residual, cross, and engineering verification, including: (1) Reliability verification, as follows: Residual analysis: calculation The residual mean is ≤0.05 and the standard deviation is ≤0.1. Cross-validation: Using the leave-one-out method, 10% of the data is removed each time during the inversion process, and the inversion accuracy is ≥90%. Engineering verification: Substitute into the numerical simulation software FLAC 3 D, The relative error between the calculated deformation value and the monitored value is ≤10%; (2) Survey report generation: includes parameter inversion results Including confidence intervals, coupling relationship graphs, data fusion quality assessment, reliability verification conclusions, and engineering application recommendations.
[0042] Example 2 This embodiment provides a method for coupled inversion of physical and mechanical parameters of soil and rock masses, applied to the investigation of a highway slope in a mountainous area. The slope is about 28m high, and the strata from top to bottom are silty clay layer and strongly weathered argillaceous sandstone layer. The project needs to obtain key physical and mechanical parameters such as elastic modulus E, Poisson's ratio μ, cohesion c, internal friction angle φ, permeability coefficient k, and porosity g for slope stability evaluation and support design.
[0043] Step 1: Collaborative Acquisition of Multi-Source Exploration Data and Establishment of a Five-Dimensional Data Ledger Five types of data were collected according to the requirements of the five-dimensional ledger, and GNSS was used to achieve unified synchronization of spatial coordinates: 1. Geological drilling data: 3 boreholes were drilled with a sampling interval of 0.8m, and natural density ρ and water content ω were obtained, totaling 126 sets; 2. In-situ test data: Standard Penetration Test (SPT) and Static Cone Penetration Test (CPT), with test depth intervals of 0.5m, totaling 213 sets; 3. Indoor test data: triaxial compression test and direct shear test, with 35 valid lithological samples per group, satisfying n≥30; 4. Geophysical data: seismic wave velocity test, resistivity test, with a survey line spacing of 4.0m, totaling 187 sets; 5. Dynamic monitoring data: Time-series monitoring data of slope surface displacement and deep settlement, totaling 720 sets.
[0044] Establish a five-dimensional data ledger that includes data type, spatial coordinates, acquisition time, testing instruments, and operators to achieve full-process traceability.
[0045] Step 2: Multi-source data preprocessing and spatiotemporal fusion 1. Outlier removal: The Z-score method is used, |Z i |>3 was identified as abnormal. A total of 19 abnormal data groups were identified and removed. The data was then filled in by linear interpolation of adjacent data. After removing each type of data, the remaining amount of each data group was greater than 50%, so no overall review was required. 2. Data normalization: Map parameters with different dimensions to the [0,1] interval to eliminate the influence of dimensional differences on the inversion; 3. Adaptive weighted spatiotemporal fusion: weight w k =0.6α k +0.4β k The reliability of the indoor test was α=0.94, the in-situ test was α=0.82, and the geophysical exploration was α=0.71; the spatiotemporal matching degree was β. k =1 / (1+d k ), d k The Euclidean distance between the collection points and the target points was used to obtain a high-quality fusion dataset of 986 sets.
[0046] Step 3: Modeling the coupling relationship between physical and mechanical parameters of soil and rock 1. Selection of coupling parameters: Determine elastic parameters (E and μ), strength parameters (c and φ), and seepage parameters. Three core coupling pairs of physical parameters (k and g); 2. Coupling coefficient calculation: Pearson correlation analysis was used to obtain C. cφ =0.83, C Eμ =0.76, C kg =0.81, all satisfying |C ij |≥0.6, is considered a strong coupling; 3. Coupled Model Construction: Obtained through linear regression fitting: c = 0.83 · φ + 4.26; E = 0.76 μ + 0.52; k = 0.81 g - 0.032.
[0047] Step 4: Construct the coupled inversion objective function Set weights ω1=0.75, ω2=0.18, =0.07, satisfying ω1+ω2+ =1, the objective function is: L( =0.75L_{fit}(\theta)+0.18L_{couple}(\theta)+0.07L_{reg}(\theta), achieving coordinated optimization of fitting accuracy, coupling constraints, and regularization.
[0048] Step 5: Set multi-dimensional inversion constraints 1. Geological constraints: Silty clay c∈[8,45] kPa, φ∈[12°,28°]; 2. Engineering experience constraint: Permeability coefficient k∈[10^ -8 ,10^ -6 ] m / s; 3. Physical constraints: Poisson's ratio μ∈[0.2,0.4], φ≥0°, k>0; 4. Dynamic monitoring constraint: The difference between the calculated displacement value and the measured value is ≤ ±2mm.
[0049] Step 6: Solve using the improved adaptive particle swarm optimization algorithm Particle swarm size N=80, maximum number of iterations T max =150; Adaptive inertial weight w(t) = 0.9 - 0.5 t / T is used. max The dynamic learning factor c1 decreases with iteration, while c2 increases; the crowding threshold in the normalized space. =0.01, to avoid premature particle convergence; global optimal change for 5 consecutive generations <1×10^ -4 That is, convergence.
[0050] Step 7: Iterative solution and dynamic update of the coupled model The initial parameter vector is the mean of the indoor experiments. Every 5 iterations, the coupling coefficients are recalculated and the coupling model is updated based on the current parameters. For parameters that do not meet the constraints, a penalty function is used to correct the fitness. After iteration until convergence, the optimal parameter vector is output. .
[0051] Step 8: Result Verification and Survey Report Generation 1. Residual verification: residual mean = 0.032 ≤ 0.05, residual standard deviation = 0.076 ≤ 0.1; 2. Cross-validation: Leave-one-out method accuracy = 93.6% ≥ 90%; 3. Engineering Validation: FLAC 3 The relative error of deformation in the numerical simulation of D is 7.2% ≤ 10%; If all three indicators are met, a survey report will be generated that includes parameter inversion results, coupling relationship maps, data fusion quality, reliability conclusions, and engineering recommendations.
[0052] The above description is only a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the concept and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for coupled inversion investigation of physical and mechanical parameters of soil and rock masses, characterized in that, Includes the following steps: Step 1: Collaborative acquisition of multi-source exploration data, determination of data types and spatiotemporal synchronous acquisition, and establishment of a five-dimensional data ledger; Step 2: Multi-source data preprocessing and spatiotemporal fusion, obtaining a unified data source through outlier removal, normalization, and adaptive weighted fusion; Step 3: Model the coupling relationship between physical and mechanical parameters of soil and rock, select coupling parameter pairs, quantify coupling coefficients and construct coupling models; Step 4: Construct the coupled inversion objective function, integrating data fitting error, coupling constraint error, and regularization term; Step 5: Set multi-dimensional inversion constraints, including geological conditions, engineering experience, physical significance, and dynamic monitoring constraints; Step 6: Design of an improved adaptive particle swarm optimization algorithm, employing adaptive inertia weights, dynamic learning factors, and crowding control; Step 7: Coupled inversion iterative solution and parameter update, iteratively optimize parameters and dynamically update the coupled model; Step 8: Verify the reliability of the inversion results and generate the survey report. Output the report through residual, cross and engineering verification.
2. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that: The data types mentioned in step 1 include geological drilling data, in-situ test data, indoor test data, geophysical data, and dynamic monitoring data; a GNSS positioning system is used to synchronize the spatial coordinates of all data, and a five-dimensional data ledger is constructed that includes data type, spatial coordinates, acquisition time, testing instruments, and operators.
3. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that: Step 2 specifically includes: (1) Outlier removal: Outliers are identified and removed using the Z-score method. The calculation formula is as follows: ； in, For the first One data value, The mean of the data. The standard deviation is denoted as ; when If a value is identified as an outlier, it is replaced by linear interpolation of adjacent data. If the remaining amount after removing a single type of data is less than 50%, the transportation data is retained and marked for review. (2) Data normalization: Normalize data of different dimensions to the interval [0,1], using the following formula: ； in, , These are the minimum and maximum values of the data, respectively. The data is after normalization; (3) Spatiotemporal fusion: An adaptive weighted fusion model is used to fuse multi-source data from the same spatial location and time period. The formula is: ； in, The merged data values; The number of data sources, 1 ≤ ≤5; For the first Weight coefficients of data sources Based on data credibility Spatiotemporal matching degree Sure, ; For data credibility, For spatiotemporal matching degree; For the first Data source in spatial coordinates ,time The data value.
4. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that: Step 3 specifically includes: (1) Screening of coupling parameters: Based on engineering mechanics theory and indoor test data, core coupling parameter pairs were screened. The screened coupling parameter pairs include: elastic parameter coupling of elastic modulus E and Poisson's ratio μ, strength parameter coupling of cohesion c and internal friction angle φ, and seepage coupling of permeability coefficient k and porosity g. Physical parameter coupling; (2) Coupling relationship quantification: Pearson correlation analysis is used to calculate and establish the coupling coefficient Cij, and the coupling correlation matrix is established. The formula is as follows: ； in, For parameters and The coupling coefficient has a value range of [-1, 1]. A value ≥0.6 indicates strong coupling; This refers to the number of samples in the indoor test. ≥30; For the first The first sample Parameter values, For the first The mean of each parameter; the coupling correlation matrix , For parameter dimensions, ≥3; (3) Coupled model construction: A parameter coupling model is established based on the coupling coefficients, as follows: The elastic parameter coupling model is: E=C Eμ ·μ+a; The strength parameter coupling model is: c=C cφ ·φ+b; seepage The physical parameter coupling model is: k=C kg ·g+e; Where a, b, and e are intercept terms, obtained by linear regression fitting.
5. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that: The coupled inversion objective function constructed in step 4 is: ； in, The parameter vector to be inverted; , , These are the weighting coefficients. Value range: ∈[0.6,0.8], ∈[0.1,0.3], ∈[0.05,0.15]; For data fitting error: , To integrate data volume, For measured fusion data, For parameters The corresponding calculated data; For coupling constraint error: , For parameters Calculated coupling coefficients; For regularization terms: This is to prevent parameter overfitting.
6. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that, The multi-dimensional constraints mentioned in step 5 include: (1) Geological constraints: Determine the range of parameter values based on the lithology of the exploration area; (2) Engineering experience constraints: Refer to the parameter range of similar projects in the same area; (3) Physical constraints: The parameter values must satisfy physical laws; (4) Dynamic monitoring constraints: The time-series deformation data satisfies , To improve monitoring accuracy.
7. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that, The improved adaptive particle swarm optimization algorithm described in step 6 includes: (1) Algorithm improvement strategy: Adaptive inertia weights: , For the number of iterations, This represents the maximum number of iterations. Dynamic learning factor: , ; Crowding control: Calculate the Euclidean distance between particles in the normalized parameter space. When the distance is less than a threshold... When the value is 0.01, the particle position is randomly reset; (2) Algorithm execution flow: Initialization: Particle swarm size N∈[50,100], parameter search space is determined by the constraints in step 5; Fitness calculation: based on the objective function This is the fitness value; Particle Update: Velocity ,Location , , A random number in the range [0,1]. For the individual's optimal, It is the global optimum; Convergence criterion: 5 consecutive generations Change < 1×10 -4 or reaching the maximum number of iterations. ,100≤ ≤200.
8. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that, Step 7 uses the mean value of the indoor experiment as the initial parameter vector. Every 5 iterations, the coupling coefficient is recalculated based on the current parameters and the coupling model is updated. The parameters obtained in each iteration are constrained and verified. The fitness value of parameters that do not meet the constraints is corrected by the penalty function method. When the convergence criterion is met, the optimal parameter vector is output.
9. The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass according to claim 1, characterized in that, Step 8's reliability verification includes residual analysis, leave-one-out cross-validation, and FLAC. 3 D. Numerical simulation engineering verification; three indicators must be met simultaneously: residual mean ≤ 0.05 and standard deviation ≤ 0.1, cross-validation accuracy ≥ 90%, and relative error between deformation calculation value and monitoring value ≤ 10%. An exploration report is generated after all indicators are met.
10. According to claim 1 The method for coupled inversion investigation of physical and mechanical parameters of soil and rock mass as described in any one of the nine claims is characterized in that, The physical and mechanical parameters of the soil and rock mass include elastic modulus E, Poisson's ratio μ, cohesion c, internal friction angle φ, permeability coefficient k, and porosity g.