Slope soil parameter random field simulation method, system, equipment and medium
By constructing an information diffusion distribution model and performing equal probability transformations, a non-Gaussian random field is generated, which solves the problem of reflecting the fluctuations in soil and rock parameters and improves the accuracy and computational efficiency of slope stability analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA THREE GORGES RENEWABLES (GRP) CO LTD
- Filing Date
- 2026-01-07
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies are unable to accurately reflect the random fluctuations of soil and rock parameters, leading to distortions in soil and rock reliability analysis. Traditional methods are inefficient and time-consuming.
By constructing an information diffusion distribution model and combining finite element software and equal probability transformation, a method for simulating slope soil parameters that conforms to a non-Gaussian random field is generated, thereby achieving efficient simulation of slope soil parameters.
It improves the accuracy of statistical estimation of slope instability probability and safety factor, realizes batch automated calculation, and improves numerical calculation efficiency.
Smart Images
Figure CN122020781A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of geotechnical engineering technology, and in particular to a method, system, equipment and medium for random field simulation of slope soil parameters. Background Technology
[0002] As a product of nature, soil and rock masses are affected by factors such as internal soil composition, sedimentary conditions, weathering degree, and human modification. As a result, the physical and mechanical parameters of soil masses exhibit varying degrees of differences, which is called spatial variability.
[0003] Existing techniques for characterizing the spatial variability of soil and rock parameters rely on classical probability distributions as probabilistic models. However, the curves obtained using classical probability distribution estimation methods are unimodal. Real soil is influenced by environmental factors and formation conditions, resulting in significant fluctuations in soil parameters. Therefore, the aforementioned methods struggle to reflect the stochastic fluctuations of soil layers. If the chosen theoretical distribution differs from the actual distribution, increasing the amount of data and computational precision cannot reduce the bias in the analysis results, leading to distortions in soil and rock reliability analysis. Commonly used deterministic analysis methods for slope stability include the Limit Equilibrium Method (LEM) and the Finite Element Method (FEM). Traditional FEM does not consider the impact of slope deformation on stability. In contrast, FEM does not require assumptions about the failure mode and location of the critical slip surface. Furthermore, a stochastic finite element method for slope reliability analysis is proposed, combining random field theory, probabilistic analysis theory, and the finite element method. This method first generates a random field model by repeatedly sampling from the probability distribution of material parameters, performs deterministic analysis using a finite element solver, and then estimates the system reliability using the discrete deterministic analysis results. However, the stochastic finite element method requires modification of the finite element source code for each random field implementation, which is time-consuming and inefficient. Summary of the Invention
[0004] To address the aforementioned technical problems, this disclosure provides a random field simulation method for slope soil parameters, including: The spatial dimension and geometric dimensions of the site to be simulated are determined based on soil parameter data. An information diffusion distribution model is constructed based on the soil parameter data and compared with multiple probability distributions to obtain a probability distribution type with high fitting degree. Based on the spatial dimension and geometric dimensions, a finite element model of the slope is established using finite element software, and finite element mesh information and node coordinates are extracted. A standard Gaussian random field is calculated based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and the standard Gaussian random field is mapped to non-Gaussian random field discrete values that conform to the probability distribution type through equal probability transformation. The non-Gaussian random field discrete values are written into an input file generated by the finite element software, and the input file is extracted using a matrix laboratory and submitted to the finite element software for finite element analysis to obtain the slope soil parameter random field simulation results.
[0005] Furthermore, the construction of the information diffusion distribution model based on the soil parameter data includes: The width of the information diffusion window is calculated based on the soil parameter data. The probability density function and cumulative distribution function are calculated using the width of the information diffusion window. An information diffusion distribution model for the spatial variability of soil parameters to be simulated is constructed using the probability density function and the cumulative distribution function.
[0006] Furthermore, the width of the information diffusion window is calculated based on the maximum and minimum values in the soil parameter data.
[0007] Furthermore, the plurality of probability distributions include: information diffusion distribution, truncated normal distribution, log-normal distribution, truncated Gumbel distribution, and Weibull distribution.
[0008] Further, the step of calculating a standard Gaussian random field based on the node coordinates and the autocorrelation function, and mapping the standard Gaussian random field to discrete values of a non-Gaussian random field conforming to the probability distribution type through an equal probability transformation, includes: Calculate the autocorrelation matrix based on the node coordinates and the autocorrelation function; Perform Cholliski decomposition on the autocorrelation matrix to obtain a lower triangular matrix; Calculate the standard Gaussian random field based on the lower triangular matrix; By mapping the standard Gaussian random field through the equal probability transformation, discrete values of a non-Gaussian random field that conform to the probability distribution type are obtained.
[0009] Further, the step of writing the discrete values of the non-Gaussian random field into an input file generated by the finite element software includes: Extract the input file containing node coordinates and element information generated by the finite element software when establishing the finite element model of the slope; The corresponding element set, cross-sectional properties, and material parameters in the input file are matched with the non-Gaussian random field discrete values; the matched non-Gaussian random field discrete values are used to replace the corresponding material property parameters in the input file to obtain an updated input file.
[0010] Furthermore, it also includes: The random field simulation results of the slope soil parameters are extracted through the interface of the matrix laboratory; The random field simulation results of the slope soil parameters are statistically processed and the statistical results are output.
[0011] This disclosure provides a random field simulation system for slope soil parameters, including: The system comprises the following modules: an acquisition module for determining the spatial dimension and geometric dimensions of the site to be simulated based on soil parameter data; a construction module for constructing an information diffusion distribution model based on the soil parameter data and comparing it with multiple probability distributions to obtain a probability distribution type with a high degree of fit; an extraction module for establishing a slope finite element model using finite element software based on the spatial dimension and geometric dimensions and extracting finite element mesh information and node coordinates; a calculation module for calculating a standard Gaussian random field based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and mapping the standard Gaussian random field to non-Gaussian random field discrete values that conform to the probability distribution type through equal probability transformation; and an output module for writing the non-Gaussian random field discrete values into an input file generated by the finite element software, extracting the input file using a matrix laboratory, submitting it to the finite element software for finite element analysis, and obtaining the slope soil parameter random field simulation results.
[0012] This disclosure provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the random field simulation method for slope soil parameters.
[0013] This disclosure provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the random field simulation method for slope soil parameters. The technical solution provided in this disclosure has the following advantages compared with the prior art: This application establishes a closed loop, enabling observed statistical characteristics to influence engineering response results end-to-end. Joint simulations of non-Gaussian margins and their correlation with the real space better reflect extreme values, truncation, or skewed behavior than simple Gaussian assumptions, thus making statistical estimates of engineering quantities such as slope instability probability and safety factors more reliable. Through a programmed data extraction, write-back, submission, and extraction process, traceable batch calculations can be achieved, facilitating engineering decision-making and risk quantification. For large-scale problems, approximation or dimensionality reduction methods can be introduced to balance accuracy and efficiency, adapting to different engineering needs and computing power conditions, achieving automated batch calculations, and improving the efficiency of numerical computation. Attached Figure Description
[0014] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.
[0015] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a schematic diagram of the random field simulation method for slope soil parameters described in the embodiments of this disclosure; Figure 2 For this disclosure Figure 1 A schematic diagram of a method for obtaining discrete values of non-Gaussian random fields; Figure 3 This is a schematic diagram comparing the information diffusion distribution described in the embodiments of this disclosure with multiple probability distributions; Figure 4 This is a schematic diagram of the random field simulation system for slope soil parameters described in the embodiments of this disclosure. Detailed Implementation
[0017] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0018] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.
[0019] Figure 1 This is a schematic diagram of the random field simulation method for slope soil parameters described in the embodiments of this disclosure; as follows: Figure 1 As shown, the random field simulation method for slope soil parameters includes: Step S1: Determine the spatial dimension and geometric dimensions of the site to be simulated based on soil parameter data. In this embodiment, the model spatial dimensions and discrete point layout determine the spatial resolution and related structure of the random field. The autocorrelation function is a function that describes the statistical correlation between two points in space. By statistically analyzing the mean, variance, coefficient of variation, and two-point autocorrelation function using measured samples (i.e., soil parameter data), the observed statistical characteristics can be mapped onto the simulation domain, ensuring that the simulation results are representative of engineering applications. Define the physical boundaries (length, width, height, or thickness) of the simulation domain and select the grid density (cell size). Pair the spatial coordinates of the measured sample points with the measured values to estimate the anisotropic correlation distance. Check whether the sample size and spatial coverage meet statistical robustness. Determine the site geometry and spatial resolution based on the field measured data to avoid scale mismatch caused by improper grid division, ensuring the physical interpretability and engineering relevance of the simulation results from the source. Provide an accurate geometric basis for subsequent autocorrelation matrix construction and cell / node mapping, reducing spatial bias when assigning parameters.
[0020] Step S2: Construct an information diffusion distribution model based on soil parameter data, and compare it with multiple probability distributions to obtain the probability distribution type with high fitting degree; In this embodiment, the information diffusion distribution is a probability distribution model constructed based on the idea of information diffusion, used to describe the marginal distribution characteristics of samples (capturing truncation, skewness, etc.). Introducing the information diffusion distribution allows for more flexible fitting of sample distributions. This includes: calculating the window width, estimating the probability density function, comparing the goodness of fit of the cumulative distribution function with other classical probability distributions. The information diffusion window width is determined by combining the minimum and maximum values of the samples with the sample size; the probability mapping requires obtaining the cumulative distribution function. Histograms and smoothing kernels are used to estimate the initial distribution shape; parameter estimation is performed on candidate distributions; multiple classical probability distributions are compared and the optimal distribution is selected. The information diffusion distribution can capture the skewness, truncation, and heavy tail characteristics of samples, avoiding the use of a single normal assumption to mask extreme / marginal behaviors. Parallel comparison with multiple candidate distributions ensures that the selected distribution is statistically superior to traditional alternative distributions. The better fit of the information diffusion distribution to tail behavior helps to more accurately estimate the sliding critical value and instability probability, directly enhancing the reliability of risk assessment. Step S3: Based on the spatial dimension and geometric dimensions, a finite element model of the slope is established using finite element software, and finite element mesh information and node coordinates are extracted; In this embodiment, the finite element model provides discrete element and node coordinate information, and random field values need to correspond one-to-one with elements or nodes. The input file (e.g., .inp) exported from the finite element software (ABAQUS) contains element sets, node coordinates, material definitions, etc., which can be read by the program and written back to the material parameter values. During the finite element modeling stage, the element type (2D / 3D), size, and meshing strategy are determined; the granularity of parameter assignment is decided—whether by element center, element as a whole, or by node; the input file is exported and the file encoding / naming specifications are guaranteed; and the parameter fields that need to be replaced or injected (such as key labels for material properties like elastic modulus, friction angle, cohesion, saturation, etc.) are recorded. A one-to-one correspondence between the random field and the engineering mesh ensures that uncertainty is directly reflected in the engineering calculation input, enhancing the feasibility of the simulation. Assignment by node or element center is supported, facilitating trade-offs between different levels of precision and computational resources. Determining the labels and locations of replacement parameters during the modeling stage reduces the risk of errors and format mismatches in subsequent file write-back. Based on structured mesh information, sample-level parallel scheduling can be more easily achieved, improving the efficiency of batch simulations. Step S4: Calculate a standard Gaussian random field based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and map the standard Gaussian random field to a non-Gaussian random field discrete value that conforms to the probability distribution type through equal probability transformation; In this embodiment, the standard Gaussian random field is a field whose edges follow a standard normal distribution but has a defined spatial correlation structure. Equal probability transformation is used to convert the Gaussian random field into a random field with arbitrary edge distributions while preserving the spatial correlation structure. The non-Gaussian random field is mapped to the target edge distribution field through equal probability transformation. A spatially correlated Gaussian field is generated through Cholesky decomposition, and the target edge distribution is strictly preserved using equal probability transformation, balancing both types of statistical properties. Preserving the spatial structure ensures that the parameter field is continuous and correlated within its neighborhood. Equal probability transformation avoids edge distribution bias caused by simple scaling transformations, making the parameter statistics (mean, variance, skewness, kurtosis) more closely match the observations. Step S5: The discrete values of the non-Gaussian random field are written into an input file generated by the finite element software. The input file is extracted using a matrix laboratory and submitted to the finite element software for finite element analysis to obtain the simulation results of the random field of slope soil parameters.
[0021] In this embodiment, discrete values of a non-Gaussian random field are injected into the material parameter domain of the finite element model, forming an engineering model with spatial variability. The finite element solver (ABAQUS) is then used to calculate the slope's deformation, stress, and stability (such as safety factors and slip surfaces). Field mapping identifies the location of material properties in the .inp file and constructs a mapping table; file writing uses MATLAB (Matrix Laboratory) combined with Python scripts to replace and generate new .inp files; automatic submission uses MATLAB to call the Python interface to submit the job to the finite element software (ABAQUS), runs the analysis, and generates result files (such as .odb); result extraction uses MATLAB's Python interface to read the .odb file and extract the required engineering quantities (displacement, stress, reaction force, safety factor, etc.). This programmed replacement, automatic submission, and result extraction form a repeatable and auditable workflow, significantly reducing manual intervention and human error. The automated process reduces manual modification and trial-and-error time, shortens the single test cycle, improves engineer efficiency, and reduces labor costs. Based on the MATLAB–Python–ABAQUS interface implementation, it has good modularity and cross-platform extensibility, and is easy to link with GIS, monitoring systems or databases.
[0022] This application establishes a closed loop, enabling observed statistical characteristics to influence engineering response results end-to-end. Joint simulations of non-Gaussian margins and their correlation with the real space better reflect extreme values, truncation, or skewed behavior than simple Gaussian assumptions, thus making statistical estimates of engineering quantities such as slope instability probability and safety factors more reliable. Through a programmed data extraction, write-back, submission, and extraction process, traceable batch calculations can be achieved, facilitating engineering decision-making and risk quantification. For large-scale problems, approximation or dimensionality reduction methods can be introduced to balance accuracy and efficiency, adapting to different engineering needs and computing power conditions, achieving automated batch calculations, and improving the efficiency of numerical computation.
[0023] In another embodiment of this application, an information diffusion distribution model is constructed based on soil parameter data, including: The width of the information diffusion window is calculated based on soil parameter data; The probability density function and cumulative distribution function are calculated using the information diffusion window width. An information diffusion distribution model of the spatial variability of soil parameters to be simulated is constructed using probability density function and cumulative distribution function.
[0024] In this embodiment, based on the soil parameter data collected in the field, the samples are first statistically described to obtain basic statistical quantities such as minimum, maximum, sample size, mean, and variance. According to information diffusion theory, one of the key control variables for the information diffusion distribution model is determined to be the information diffusion window width. The window width is used to control the smooth diffusion degree of information within the sample interval and the local morphology of the probability density. The probability density function and cumulative distribution function are constructed by accumulating and normalizing local information on the sample interval using the window width as the scale. The specific process includes sample preprocessing, window width estimation, local kernel density accumulation, normalization to obtain the probability density function, and integration to obtain the cumulative distribution function. The obtained probability density function and cumulative distribution function describe the statistical characteristics of the soil parameters to be simulated on the sample interval, thereby constructing an information diffusion distribution model, which serves as one of the candidate models for the marginal distribution of parametric random fields. The information diffusion distribution introduces the local adjustment capability of the window width when constructing the probability density, enabling more flexible characterization of the skewness, truncation, and kurtosis characteristics of the samples. Especially when there are outliers in the samples or the sample distribution is asymmetric, it can provide a description that fits the observation data more closely than a traditional single distribution. By accumulating local statistics across the window width, biases arising from relying solely on parameter estimation are effectively avoided, thus improving the robustness and reliability of the marginal distribution fitting. An invertible cumulative distribution function is provided for subsequent equal-probability transformations, ensuring the numerical feasibility and precision control of the mapping process from a Gaussian field to the target non-Gaussian field, facilitating engineering implementation and software application.
[0025] In another embodiment of this application, the information diffusion window width is calculated based on the maximum and minimum values in the soil parameter data.
[0026] In this embodiment, given the sample size (soil parameter data) and sample interval, the window width can be estimated using empirical or theoretical formulas based on the sample interval range. Specifically, the difference between the maximum and minimum values of the samples is used as the overall scale reference, and the window width is determined by combining the influence coefficient of the sample size. The calculation of the window width considers both the total span of the sample interval and the adjustment of local smoothness through weights that vary with the sample size, ensuring that the window width is wider when the sample is sparse to smooth noise and narrower when the sample is abundant to preserve details. The window width value should be provided in the specification with corresponding empirical formulas or parameter tables, along with an explanation of the numerical solution or interpolation method. Calculating the window width directly using sample extreme values is a simple and easy method with clear engineering interpretability, does not rely on excessive hyperparameter settings, and facilitates rapid implementation on-site. A window width based on the sample interval can adapt to both sparse and dense sample conditions, improving the robustness of probability density estimation, thus enabling the subsequent random field marginal distribution fitting results to have a certain degree of adaptability to sample size fluctuations.
[0027] In another embodiment of this application, the plurality of probability distributions include: information diffusion distribution, truncated normal distribution, log-normal distribution, truncated Gumbel distribution, and Weibull distribution.
[0028] In this embodiment, to ensure that the selected marginal distribution reflects both sample characteristics and engineering interpretability, the information diffusion distribution and several commonly used distributions are considered as candidate options. The candidate distributions include the truncated normal distribution, log-normal distribution, truncated Gumbel distribution, and Weibull distribution. For each candidate distribution, parameters are determined through maximum likelihood estimation or moment estimation, and multiple goodness-of-fit evaluation indices are used to compare and select the optimal model. It is recommended that commonly used statistics for distribution fitting be used as goodness-of-fit evaluation indices, such as the maximum difference between the empirical and theoretical distributions, the information criterion, and the root mean square error, supplemented by a visual comparison of histograms and fitted curves to ensure that the selected distribution is both statistically significant and engineering interpretable. Placing the information diffusion distribution in the candidate set with traditional distributions allows for statistical comparison and selection, further ensuring the rationality of the marginal distribution and the objectivity of the judgment. The multi-model comparison mechanism reduces the systematic bias caused by a single distribution assumption, especially in marginal event estimation and tail probability calculation, significantly improving the accuracy of the judgment. In another embodiment of this application, Figure 2 For this disclosure Figure 1 A schematic diagram illustrating methods for obtaining discrete values of non-Gaussian random fields; as shown. Figure 2 As shown, in step S4, a standard Gaussian random field is calculated based on the node coordinates and autocorrelation function, and then mapped to a non-Gaussian random field discrete value conforming to the probability distribution type through an equal probability transformation, including: Step S41: Calculate the autocorrelation matrix based on the node coordinates and the autocorrelation function; Step S42: Perform Cholliski decomposition on the autocorrelation matrix to obtain the lower triangular matrix; Step S43: Calculate the standard Gaussian random field based on the lower triangular matrix; Step S44: Map the standard Gaussian random field through equal probability transformation to obtain discrete values of a non-Gaussian random field that conform to the probability distribution type.
[0029] In this embodiment, the first stage is the construction of spatial correlation. Based on the node coordinates exported by the finite element software, the correlation coefficients between node pairs are calculated according to a pre-determined autocorrelation function model, thus constructing the autocorrelation matrix. To ensure that the autocorrelation matrix can be used for generating correlated samples, it should be ensured that the matrix is a positive definite matrix. If necessary, a small perturbation term can be added to the diagonal to meet the numerical stability requirements. Subsequently, the autocorrelation matrix is decomposed by Cholesky to obtain a lower triangular matrix. By generating independent and identically distributed standard normal random vectors and multiplying them with the lower triangular matrix, a standard Gaussian random field with the desired spatial correlation structure can be obtained. The second stage is the marginal distribution permutation. Each component of the standard Gaussian random field is mapped to a probability value between zero and one through a standard normal cumulative distribution. Then, the probability value is inversely transformed using the inverse function of the previously determined target marginal distribution, thereby obtaining a non-Gaussian random field discrete value that conforms to the target marginal distribution and retains spatial correlation. The standard Gaussian random field generated by Cholesky decomposition can strictly recover the spatial correlation structure defined by the autocorrelation function, ensuring that the parameter field has reasonable continuity and correlation in space, and avoiding isolated noise values without physical meaning. By using an equal probability transformation to transform the marginal distribution from Gaussian to the target distribution, the final parameter field satisfies both spatial correlation and marginal statistical characteristics, achieving dual consistency from a statistical perspective and significantly improving the representativeness of the parameter field in engineering simulation.
[0030] In another embodiment of this application, writing the discrete values of the non-Gaussian random field into an input file generated by finite element software includes: Extract the input file containing node coordinates and element information generated by the finite element software when building the finite element model of the slope; The corresponding element set, section properties, and material parameters in the input file are matched with the non-Gaussian random field discrete values; the matched non-Gaussian random field discrete values are then used to replace the corresponding material property parameters in the input file to obtain the updated input file.
[0031] In this embodiment, the discrete values of the non-Gaussian random field are mapped to the data structure of the finite element model and written into the input file generated by the finite element software. This includes extracting the input file containing node coordinates and element information generated when establishing the slope finite element model, and parsing the identifiers and locations of element sets, section properties, and material parameters in the input file. According to a predetermined mapping rule, the discrete values of the random field are matched with the corresponding element sets or node numbers in the input file, and then the corresponding material property parameters in the input file are replaced with the matched discrete values to generate an updated input file. The updated input file can then be used for calculation. By establishing a mapping relationship between the fields of the input file and the discrete values of the random field, accurate parameter injection can be achieved, ensuring that the material parameters used in the finite element model truly reflect spatial variability. The automated replacement process eliminates the tediousness and error-proneness of manual item-by-item modification, significantly reducing inconsistencies caused by human operation, improving work efficiency, and reducing the error rate. The updated input file is directly used for solving, shortening the cycle time from parameter generation to engineering solution.
[0032] In another embodiment of this application, it further includes: Extract the random field simulation results of slope soil parameters through the interface of the matrix laboratory; The results of random field simulation of slope soil parameters are statistically processed and output.
[0033] In this embodiment, the interface is a program interface provided by the matrix laboratory for data transmission and result manipulation with the finite element software. The interface extracts the result files generated by the finite element software into the matrix laboratory, analyzes the required response quantities such as displacement, stress, strain, or safety factor, and then performs statistical processing on the results of multiple samples, including calculating the mean, variance, confidence interval, probability density estimation, and instability probability estimation. It can also output result distribution diagrams, sensitivity ranking tables, and engineering criteria (such as minimum safety factor, maximum displacement, etc.) as needed. The interface enables data exchange between the solver and the matrix laboratory, automating the acquisition and centralized management of simulation results, facilitating the unified processing of batch simulation results.
[0034] In another embodiment of this application, the random field simulation process for slope soil parameters is as follows: The correlation distance and autocorrelation function of the parameters to be simulated are: ; ; In the formula: δ is the correlation distance, ρ(τ) is the correlation function, and τ is the relative distance between the two points. and These represent the differences in coordinates between two points in space along the x and y directions, respectively. and These are the relevant distances in the horizontal and vertical directions, respectively.
[0035] The probability density function f(x) and cumulative distribution function F(x) of the information diffusion distribution model are as follows: ; ; In the formula: g and b are the maximum and minimum values in the sample, and d is the window width during information diffusion. Please solve. The value of is related to the observed quantity n, as shown in Table 1.
[0036] Table 1
[0037] Figure 3 This is a schematic diagram comparing the information diffusion distribution described in the embodiments of this disclosure with multiple probability distributions; as shown Figure 3 As shown, frequency distribution histograms of the information diffusion distribution and the classical probability distribution are plotted, and probability density function curves are drawn. The curve with the highest goodness of fit is selected as the probability distribution function type for the parameters. These include the information diffusion distribution, truncated normal distribution, log-normal distribution, truncated Gumbel distribution, and Weibull distribution. Based on the image fitting effect, the information diffusion distribution with a higher goodness of fit is selected.
[0038] Determine the dimensions and geometric size of the site to be simulated; since maintenance roads are mostly built in mountains, the road slopes are usually trapezoidal; the corresponding fourth step is to establish an ABAQUS finite element model of the simulated site based on the actual road slope size, extract the finite element mesh information, and selectively extract data points from the rule model according to actual needs during the random field modeling process.
[0039] The autocorrelation matrix is calculated based on the coordinates of the element center point and the autocorrelation function, and then Cholesky decomposition is performed on the autocorrelation matrix. First, to characterize the random field of soil parameters, the centroids of the random field elements are discretized to establish the autocorrelation matrix C: ; The autocorrelation matrix C is decomposed into the product of a lower triangular matrix L and its transpose LT using the Choleski decomposition: ; The formula for determining the Gaussian random field model is: ; In the formula: For a standard Gaussian random field, Let be an n-dimensional standard normal random variable.
[0040] The formula for converting a standard Gaussian random field into a non-Gaussian random field that conforms to the information diffusion distribution using the equal probability transformation method is as follows: ; In the formula: Let the input parameters be the inverse function of the cumulative distribution of the random field. This is the cumulative distribution function of a standard normal variable.
[0041] Finally, data interaction was achieved through Matlab, Python, and ABAQUS. MATLAB was mainly used for probabilistic analysis of soil parameters, extraction of finite element mesh parameters, and random field simulation; ABAQUS was mainly used for finite element division of slope models and slope stability calculation; while Python acted as a connector to enable data interaction between MATLAB and ABAQUS.
[0042] Figure 4 This is a schematic diagram of the random field simulation system for slope soil parameters described in the embodiments of this disclosure, as shown below. Figure 4 As shown, this application also provides a random field simulation system for slope soil parameters, including: The module 601 is used to determine the spatial dimension and geometric dimensions of the site to be simulated based on soil parameter data; the module 602 is used to construct an information diffusion distribution model based on soil parameter data and compare it with multiple probability distributions to obtain a probability distribution type with a high degree of fit; the extraction module 603 is used to establish a slope finite element model using finite element software based on the spatial dimension and geometric dimensions and extract finite element mesh information and node coordinates; the calculation module 604 is used to calculate a standard Gaussian random field based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and to map the standard Gaussian random field to a non-Gaussian random field discrete value that conforms to the probability distribution type through equal probability transformation; the output module 605 is used to write the non-Gaussian random field discrete value into an input file generated by the finite element software, extract the input file using a matrix laboratory, and submit it to the finite element software to perform finite element analysis to obtain the slope soil parameter random field simulation results.
[0043] This application establishes a closed loop, enabling observed statistical characteristics to influence engineering response results end-to-end. Joint simulations of non-Gaussian margins and their correlation with the real space better reflect extreme values, truncation, or skewed behavior than simple Gaussian assumptions, thus making statistical estimates of engineering quantities such as slope instability probability and safety factors more reliable. Through a programmed data extraction, write-back, submission, and extraction process, traceable batch calculations can be achieved, facilitating engineering decision-making and risk quantification. For large-scale problems, approximation or dimensionality reduction methods can be introduced to balance accuracy and efficiency, adapting to different engineering needs and computing power conditions, achieving automated batch calculations, and improving the efficiency of numerical computation.
[0044] This application also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of a random field simulation method for slope soil parameters.
[0045] This application also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of a random field simulation method for slope soil parameters.
[0046] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0047] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A random field simulation method for slope soil parameters, characterized in that, include: The spatial dimension and geometric dimensions of the site to be simulated are determined based on soil parameter data. An information diffusion distribution model is constructed based on the soil parameter data and compared with multiple probability distributions to obtain a probability distribution type with high fitting degree. Based on the spatial dimension and geometric dimensions, a finite element model of the slope is established using finite element software, and finite element mesh information and node coordinates are extracted. A standard Gaussian random field is calculated based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and the standard Gaussian random field is mapped to non-Gaussian random field discrete values that conform to the probability distribution type through equal probability transformation. The non-Gaussian random field discrete values are written into an input file generated by the finite element software, and the input file is extracted using a matrix laboratory and submitted to the finite element software for finite element analysis to obtain the slope soil parameter random field simulation results.
2. The random field simulation method for slope soil parameters according to claim 1, characterized in that, The construction of the information diffusion distribution model based on the soil parameter data includes: The width of the information diffusion window is calculated based on the soil parameter data. The probability density function and cumulative distribution function are calculated using the width of the information diffusion window. An information diffusion distribution model for the spatial variability of soil parameters to be simulated is constructed using the probability density function and the cumulative distribution function.
3. The random field simulation method for slope soil parameters according to claim 2, characterized in that, The width of the information diffusion window is calculated based on the maximum and minimum values in the soil parameter data.
4. The random field simulation method for slope soil parameters according to claim 1, characterized in that, The plurality of probability distributions include: information diffusion distribution, truncated normal distribution, log-normal distribution, truncated Gumbel distribution, and Weibull distribution.
5. The random field simulation method for slope soil parameters according to claim 1, characterized in that, The step of calculating a standard Gaussian random field based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and mapping the standard Gaussian random field to a non-Gaussian random field discrete value conforming to the probability distribution type through equal probability transformation, includes: Calculate the autocorrelation matrix based on the node coordinates and the autocorrelation function; Perform Cholliski decomposition on the autocorrelation matrix to obtain a lower triangular matrix; Calculate the standard Gaussian random field based on the lower triangular matrix; By mapping the standard Gaussian random field through the equal probability transformation, discrete values of a non-Gaussian random field that conform to the probability distribution type are obtained.
6. The random field simulation method for slope soil parameters according to claim 1, characterized in that, The step of writing the discrete values of the non-Gaussian random field into an input file generated by finite element software includes: Extract the input file containing node coordinates and element information generated by the finite element software when establishing the finite element model of the slope; The corresponding element set, cross-sectional properties, and material parameters in the input file are matched with the non-Gaussian random field discrete values; the matched non-Gaussian random field discrete values are used to replace the corresponding material property parameters in the input file to obtain an updated input file.
7. The random field simulation method for slope soil parameters according to claim 1, characterized in that, Also includes: The random field simulation results of the slope soil parameters are extracted through the interface of the matrix laboratory; The random field simulation results of the slope soil parameters are statistically processed and the statistical results are output.
8. A random field simulation system for slope soil parameters, characterized in that, include: The acquisition module is used to determine the spatial dimension and geometric dimensions of the site to be simulated based on the soil parameter data; the construction module is used to construct an information diffusion distribution model based on the soil parameter data, and compare it with multiple probability distributions to obtain a probability distribution type with a high degree of fit. The extraction module is used to establish a finite element model of the slope using finite element software based on the spatial dimension and geometric dimensions, and to extract finite element mesh information and node coordinates; the calculation module is used to calculate a standard Gaussian random field based on the autocorrelation function corresponding to the node coordinates and the soil parameter data, and to map the standard Gaussian random field into a non-Gaussian random field discrete value that conforms to the probability distribution type through equal probability transformation. The output module is used to write the discrete values of the non-Gaussian random field into an input file generated by the finite element software, extract the input file using the matrix laboratory, and submit it to the finite element software to perform finite element analysis, thereby obtaining the random field simulation results of the slope soil parameters.
9. A computer device, characterized in that, It includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the random field simulation method for slope soil parameters as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the random field simulation method for slope soil parameters as described in any one of claims 1 to 7.