Unsaturated soil slope reliability analysis method based on GeoStudio intelligent agent model

Abstract

The invention discloses an unsaturated soil slope reliability analysis method based on a GeoStudio intelligent agent model, and relates to the technical field of geotechnical engineering. Acquiring a plurality of samples of soil-water characteristic curve model parameters corresponding to the unsaturated soil slope; according to the slope measurement data and the soil-water characteristic curve model, constructing a slope stability original analysis model on GeoStudio software, and exporting the slope stability original analysis model as a modifiable text file; dividing the plurality of samples into two groups, controlling GeoStudio software to automatically run a text file through a programming language script, and calculating the slope safety coefficient corresponding to the first group of samples in batches; constructing an agent model through the first group of samples and the corresponding slope safety coefficients, and respectively calculating the slope safety coefficients corresponding to the second group of samples through the agent model; and determining the slope instability probability of the unsaturated soil slope according to the slope safety coefficients corresponding to all the samples. The method can significantly improve the analysis efficiency of the slope reliability.

Description

Technical Field

[0001] The invention relates to the technical field of engineering geology and geotechnical engineering, and in particular to a reliability analysis method for an unsaturated soil slope based on a GeoStudio intelligent proxy model. Background Art

[0002] Landslide disasters are one of the most serious geological disasters in the world. Due to their suddenness and destructive power, they have become a major problem threatening human life, property and ecological environment. With global climate change, frequent extreme weather and intensified human engineering activities, the frequency and impact range of landslide disasters have increased significantly, especially in areas with complex geological structures and changeable climates, which have repeatedly caused major disasters. Therefore, the study of slope reliability has important practical significance.

[0003] In the prior art, slope reliability analysis is usually based on deterministic analysis methods, which rely on a single soil-water characteristic curve (SWCC) model or a fixed parameter combination to calculate the slope, without considering the sensitivity of SWCC to slope stability, and ignoring the randomness and uncertainty of soil properties and environmental conditions. In addition, since traditional analysis methods are mostly manually set and calculated under a single working condition, they cannot effectively cope with multi-scenario analysis under a large number of SWCC parameter combinations, resulting in low efficiency in slope reliability analysis. Summary of the invention

[0004] Based on this, it is necessary to provide a reliability analysis method for unsaturated soil slopes based on the GeoStudio intelligent proxy model to address the above technical problems. This method can improve the efficiency of slope reliability analysis.

[0005] The present invention adopts the following technical solutions:

[0006] The present invention provides a reliability analysis method for unsaturated soil slope based on GeoStudio intelligent proxy model, comprising:

[0007] Combine the slope measurement data of the unsaturated soil slope with the existing data to determine multiple samples of the soil-water characteristic curve model; the existing data are the measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope;

[0008] According to the slope measurement data and soil-water characteristic curve model, the original slope stability analysis model is constructed on GeoStudio software, and the original slope stability analysis model is exported as a modifiable text file;

[0009] Divide multiple samples into two groups, control GeoStudio software to automatically run text files through programming language scripts, and batch calculate the slope safety factors corresponding to the first group of samples;

[0010] A proxy model is constructed by using the first group of samples and the corresponding slope safety factors, and the slope safety factors corresponding to the second group of samples are calculated by using the proxy model;

[0011] The slope instability probability of the unsaturated soil slope is determined according to the slope safety factor corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

[0012] Optionally, the programming language script is used to control GeoStudio software to automatically run the text file to batch calculate the slope safety factor corresponding to the first group of samples, including:

[0013] Assigning the first group of samples to the model parameters of the soil-water characteristic curve model in the text file through a programming language script to obtain multiple model files;

[0014] The software library corresponding to the programming language controls the GeoStudio software to run each model file in turn to obtain the slope safety factor corresponding to the first group of samples.

[0015] Optionally, the GeoStudio software is controlled by the software library corresponding to the programming language to run each model file in sequence to obtain the slope safety factor corresponding to the first group of samples, including:

[0016] For any sample in the first group of samples, the software library corresponding to the programming language is used to control the GeoStudio software to run the model file corresponding to the sample, and the safety factors of multiple sliding surfaces are obtained;

[0017] The minimum safety factor among the safety factors of multiple sliding surfaces is determined as the slope safety factor corresponding to the sample.

[0018] Optionally, a proxy model is constructed using the first set of samples and corresponding slope safety factors, including:

[0019] The first group of samples and the corresponding slope safety factors are divided to obtain a training data set and a test data set;

[0020] The proxy model is constructed through the training data set, and the accuracy of the proxy model is verified through the test data set.

[0021] Optionally, the proxy model includes a polynomial proxy model; the proxy model is constructed by training the dataset, including:

[0022] Fitting a second-order polynomial without cross terms to the training data set to determine the coefficients of the second-order polynomial without cross terms;

[0023] The fitted second-order polynomial without cross terms was determined as the polynomial surrogate model.

[0024] Optionally, the proxy model includes a machine learning proxy model; the proxy model is constructed by training the dataset, including:

[0025] Based on the training data set, a machine learning agent model is constructed using the machine learning random forest algorithm.

[0026] Optionally, the slope instability probability of the unsaturated soil slope is determined based on the slope safety factors corresponding to all samples, including:

[0027] Count the number of slope safety factors less than a preset safety threshold among all slope safety factors;

[0028] The ratio of the number to the total number of all samples was determined as the slope failure probability of the unsaturated soil slope.

[0029] The present invention provides a reliability analysis device for an unsaturated soil slope based on a GeoStudio intelligent proxy model, comprising:

[0030] A determination module is used to combine the slope measurement data of the unsaturated soil slope with the existing data to determine multiple samples of soil-water characteristic curve model parameters; the existing data are measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope;

[0031] The export module is used to construct the original analysis model of slope stability on GeoStudio software according to the slope measurement data and soil-water characteristic curve model, and export the original analysis model of slope stability as a modifiable text file;

[0032] The linkage module is used to divide multiple samples into two groups, control the GeoStudio software to automatically run the text file through the programming language script, and batch calculate the slope safety factor corresponding to the first group of samples;

[0033] A construction module is used to construct a proxy model through the first group of samples and the corresponding slope safety factors, and calculate the slope safety factors corresponding to the second group of samples through the proxy model;

[0034] The calculation module is used to determine the slope instability probability of the unsaturated soil slope according to the slope safety factors corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

[0035] The present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the reliability analysis method of an unsaturated soil slope based on a GeoStudio intelligent proxy model is implemented.

[0036] The present invention provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the reliability analysis method of an unsaturated soil slope based on a GeoStudio intelligent proxy model when executing the program.

[0037] At least one of the above technical solutions adopted by the present invention can achieve the following beneficial effects:

[0038] In the present invention, an original analysis model of slope stability is constructed by GeoStudio software used for engineering geology and geotechnical engineering, and a programming language script is used to control the GeoStudio software to run automatically, so as to complete batch calculation of slope safety factors of a first group of samples, thereby improving the calculation efficiency of slope reliability analysis, and based on the first group of samples with corresponding slope safety factors, an agent model for quickly calculating slope safety factors is constructed, thereby further improving the calculation efficiency of slope reliability analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0040] Figure 1 A flow chart of a reliability analysis method for an unsaturated soil slope based on a GeoStudio intelligent proxy model provided by the present invention;

[0041] Figure 2 A schematic flow chart of another reliability analysis method for unsaturated soil slopes based on the GeoStudio intelligent proxy model provided by the present invention;

[0042] Figure 3 A schematic diagram of an automated slope reliability analysis using a Python script provided by the present invention;

[0043] Figure 4 A slope reliability analysis model diagram in an embodiment provided by the present invention;

[0044] Figure 5 A large number of slope SWCC parameter sample maps obtained based on hierarchical Bayesian provided by the present invention;

[0045] Figure 6 A proxy model data verification diagram provided by the present invention, constructed based on Python and Geo-studio linkage calculation results;

[0046] Figure 7A slope safety factor histogram calculated based on Python and Geo-stduio linkage and proxy model provided by the present invention;

[0047] Figure 8 A schematic diagram of a reliability analysis device for an unsaturated soil slope based on a GeoStudio intelligent proxy model provided by the present invention;

[0048] Fig. 9 A schematic diagram of a computer device for implementing a reliability analysis method for an unsaturated soil slope based on a GeoStudio intelligent proxy model provided by the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0050] Slope reliability analysis is an important research topic in the fields of engineering geology, geotechnical engineering and civil engineering, and is directly related to the safety and reliability of the project. In the traditional technical framework, slope reliability analysis mostly adopts a deterministic method, which assumes that the physical and mechanical parameters of the soil (such as shear strength, pore water pressure parameters, etc.) are fixed values, and combines the classical limit equilibrium method or finite element method to calculate the stability factor of the slope. However, this deterministic analysis ignores the randomness and uncertainty of soil parameters and environmental conditions, and it is difficult to fully reflect the instability risk of the slope under complex geological conditions. In addition, most of them do not consider the influence of the uncertainty of SWCC on the stability of the slope. In actual engineering, the hydraulic characteristics of unsaturated soil have a significant impact on the stability of the slope, and the parameters of SWCC usually show significant randomness. For example, due to changes in construction technology, soil formation and external environment, the SWCC parameters of different slopes or different areas of the same slope may vary significantly. The traditional analysis method cannot systematically consider these uncertainties, which may lead to biased results, thus limiting its applicability in complex engineering.

[0051] Therefore, in actual engineering, since the soil-water characteristic curve of the slope is significantly random due to the influence of multiple factors, traditional deterministic analysis is difficult to fully reflect the real slope instability risk under complex working conditions. At the same time, since traditional analysis methods are mostly manual settings and single working condition calculations, they cannot effectively cope with multi-scenario analysis under a large number of SWCC parameter combinations, resulting in low analysis efficiency, cumbersome operations, and difficulty in providing sufficiently accurate instability probability assessment.

[0052] At the same time, there are relatively few studies on the effect of SWCC on slope stability. Most studies focus more on the impact of changes in traditional soil parameters (such as cohesion and internal friction angle) on slope stability, while there is little systematic research on the probability of slope instability under different SWCC conditions.

[0053] In recent years, some scholars have tried to combine probabilistic analysis methods with traditional slope stability calculations. On the one hand, the accuracy of the prior distribution is improved based on empirical knowledge, and the posterior estimation of SWCC parameters is completed using the Bayesian method; on the other hand, the SWCC parameters of extremely sparse observation sites are estimated using the hierarchical Bayesian method based on the similarity between sites. Both methods capture the randomness and uncertainty of SWCC by introducing statistical models, but because the models are complex and have no analytical solutions, they are usually combined with Monte Carlo simulation or Latin hypercube sampling techniques to generate a large number of different SWCC posterior parameter combinations. This technological advancement provides new possibilities for further studying the impact of SWCC variability on slope stability. By analyzing these parameter combinations one by one and combining them with professional numerical simulation software such as Geo-Studio, the safety factor of the slope can be calculated and the probability of slope instability can be statistically analyzed.

[0054] Based on the above analysis, it can be determined that the existing technical solutions have many shortcomings. On the one hand, the generation and analysis of multiple sets of random parameters still need to be completed manually, which is inefficient and cumbersome. On the other hand, the lack of relevant linkage between different software leads to a lack of automated tool support for the extraction and statistics of results, especially in large-scale parameter combination analysis, which is complicated to operate and prone to data errors. In addition, some studies have attempted to analyze multiple scenarios by manually adjusting SWCC parameters, but their efficiency and accuracy are greatly limited, making it difficult to meet the needs of rapid reliability assessment under complex working conditions.

[0055] Based on this, the present invention provides a reliability analysis method for unsaturated soil slopes based on GeoStudio intelligent proxy model. The present invention is an efficient and automated analysis method that overcomes the problems of traditional manual generation of random parameters, one-by-one analysis, and complex and cumbersome result statistics. With the help of a large number of SWCC parameter posterior samples obtained by Bayesian method research, the present invention realizes batch analysis of SWCC parameter uncertainty by constructing Python and Geo-Studio linkage calculation software. At the same time, a fast calculation proxy model is constructed, and automated parameter input, calculation and result extraction tools are developed to improve analysis efficiency, avoid the inefficiency and human errors of manual operations, thereby significantly improving the accuracy and efficiency of large-scale parameter combination analysis. In addition, combined with the probability analysis method, the instability probability of the slope under a large number of SWCC conditions is calculated for rapid quantitative evaluation, providing a scientific basis for slope reliability analysis. Through the technical solution of the present invention, it is possible to fill the shortcomings of the existing technology in automated processing and efficient statistics, comprehensively consider the uncertainty of SWCC, improve the accuracy and reliability of slope reliability analysis under complex working conditions, and provide more scientific support for slope design, governance and risk assessment.

[0056] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0057] Figure 1 The figure is a flow chart of a reliability analysis method for an unsaturated soil slope based on the GeoStudio intelligent proxy model in the present invention, which specifically includes the following steps:

[0058] S101, combining the slope measurement data of the unsaturated soil slope with the existing data to determine multiple samples of soil-water characteristic curve model parameters; the existing data are measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope.

[0059] The slope measurement data of the unsaturated soil slope is combined with the existing data to determine multiple samples of the soil-water characteristic curve model parameters, including: completing the measurement of sparse data points of the slope (slope measurement data of the unsaturated soil slope) through indoor experiments, and consulting literature and engineering records to obtain existing data of sites with similar geological history and sedimentary environment and the same soil type; constructing a priori distribution based on experience and literature records, considering the slope measurement data and existing data to construct a likelihood function, and solving it using the hierarchical Bayesian method; due to the complexity of model construction, Markov chain Monte Carlo is used to complete the sampling of SWCC parameter posterior samples, and the parameter posterior samples Θ = {α i , m i , n i , i=1,2,…,N s}, the parameter posterior samples are determined as multiple samples, including N sEach sample corresponds to a soil-water characteristic curve model.

[0060] S102, constructing an original analysis model of slope stability on GeoStudio software according to the slope measurement data and the soil-water characteristic curve model, and exporting the original analysis model of slope stability as a modifiable text file.

[0061] Among them, Geo-studio software is a numerical simulation software used for engineering geology and geotechnical engineering. The slope measurement data may include soil property parameter values. Therefore, according to the slope measurement data and the soil-water characteristic curve model, the original analysis model of slope stability is constructed on the GeoStudio software, including: according to the engineering case, soil property parameter values ​​and soil-water characteristic curve model, the SLOPE / W module in the Geo-studio software inputs the geometric shape, assigns the model material value, boundary conditions and sliding in and out area, and obtains the original analysis model of slope stability.

[0062] Export the original analysis model of slope stability as a text file, including: export the original analysis model of slope stability as a modifiable text file "#.xml".

[0063] S103, dividing the multiple samples into two groups, controlling GeoStudio software to automatically run the text file through a programming language script, and batch calculating the slope safety factors corresponding to the first group of samples.

[0064] Replace different samples of text files by encoding.

[0065] Optionally, the GeoStudio software is controlled by a programming language script to automatically run the text file and batch calculate the slope safety factor corresponding to the first group of samples, including: assigning the first group of samples to the model parameters of the soil-water characteristic curve model in the text file through the programming language script to obtain multiple model files; controlling the GeoStudio software to run each model file in turn through the software library corresponding to the programming language to obtain the slope safety factor corresponding to the first group of samples.

[0066] Specifically, the programming language can be Python. For the first set of samples Θ i , use Python coding to reassign the model parameters in the text file "#.xml" in S102 to the first set of samples Θ i , generate the "i-th.xml" model file corresponding to each sample in the first group of samples. This model file represents the slope stability analysis model established by the i-th soil-water characteristic curve model; use the Autoit library in python to code and control Geo-Studio, and run different samples (α i , m i, n i ) corresponding to the "i-th.xml" file, and perform slope stability analysis corresponding to the slope stability analysis model. The slope stability analysis corresponding to a slope stability analysis model is characterized by a deterministic slope safety factor.

[0067] Optionally, the GeoStudio software is controlled by the software library corresponding to the programming language to run each model file in sequence to obtain the slope safety factor corresponding to the first group of samples, including: for any sample in the first group of samples, the GeoStudio software is controlled by the software library corresponding to the programming language to run the model file corresponding to the sample to obtain the safety factors of multiple sliding surfaces; the minimum safety factor among the safety factors of the multiple sliding surfaces is determined as the slope safety factor corresponding to the sample.

[0068] A model file represents the slope stability analysis model established corresponding to the soil-water characteristic curve model. For any model file corresponding to a sample, since the slope stability analysis model corresponds to multiple sliding surfaces, different sliding surfaces correspond to different safety factors. The minimum value of the safety factor corresponding to all sliding surfaces is extracted as the sample (α i , m i , n i ) corresponding to the slope safety factor (F t,i ), where the number of samples in the first group is N t indivual.

[0069] The present invention utilizes the proposed Python and Geo-Studio linkage program to t = 20,000 representative Markov Chain Monte Carlo (MCMC) samples The process of calculating the slope safety factor takes about 80 hours. The workstation configuration used is as follows: Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz (up to 3.50GHz), memory is 256GB.

[0070] S104, constructing a proxy model through the first group of samples and the corresponding slope safety factors, and calculating the slope safety factors corresponding to the second group of samples through the proxy model.

[0071] In slope reliability analysis, since a large number of samples are involved, whether using numerical simulation or other methods, it will take a lot of time. Therefore, it is further considered to establish a proxy model of output response. The proxy model only needs to be trained with a small number of samples to achieve fast approximation, thereby greatly improving the calculation efficiency and reducing the calculation complexity. It has a wide range of applications in large-scale sample analysis and complex model calculation. Therefore, a proxy model can be constructed based on the above first group of samples and their corresponding slope safety factors.

[0072] A proxy model is constructed through the first group of samples and the corresponding slope safety factors, including: dividing the first group of samples and the corresponding slope safety factors to obtain a training data set and a test data set; constructing the proxy model through the training data set, and verifying the accuracy of the proxy model through the test data set.

[0073] The different samples in S103 and the corresponding slope safety factor data set X = {(α i , m i , n i ), F t,i}, divided into training data set X r And the test dataset X q ; Then build a proxy model based on the training dataset and use the test dataset X q Perform performance test on the proxy model, for example, the accuracy of the proxy model reaches a preset accuracy threshold; if the accuracy of the proxy model reaches the preset accuracy threshold, the constructed proxy model is determined as the final proxy model; if the accuracy of the proxy model does not reach the preset accuracy threshold, the training data set samples can be increased until the accuracy of the proxy model reaches the preset accuracy threshold.

[0074] Optionally, the proxy model includes a polynomial proxy model; constructing the proxy model through a training data set includes: fitting a second-order polynomial without cross-terms to the training data set to determine the coefficients of the second-order polynomial without cross-terms; and determining the fitted second-order polynomial without cross-terms as the polynomial proxy model.

[0075] Among them, the second-order polynomial without cross terms is:

[0076]

[0077] Among them, F s (X) is the slope safety factor, is the coefficient of the required solution of the second-order polynomial without cross terms, is the sample, n w is the number of samples in the training data set. n w samples, corresponding to 2n w +1 polynomial unknown parameter, the solution needs to be based on a series of sampling points to fit a response surface to approximate the true limit state surface.

[0078] Optionally, the proxy model includes a machine learning proxy model; constructing the proxy model through a training data set includes: based on the training data set, constructing the machine learning proxy model through a machine learning random forest algorithm.

[0079] The slope safety factors corresponding to the second group of samples are calculated respectively through the proxy model, including: inputting each sample in the second group of samples into the proxy model respectively to obtain the slope safety factor corresponding to each sample in the second group of samples.

[0080] S105, determining the slope instability probability of the unsaturated soil slope according to the slope safety factors corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

[0081] The probability of slope instability can transform the traditional deterministic problem, that is, only looking at the slope safety factor, into an uncertain problem, so as to better provide scientific countermeasures for slope engineering design and safety maintenance. The size of the failure probability can determine the degree of slope protection. When the value is small, the protection degree is low, and when the value is large, the protection degree is high.

[0082] For the slope reliability problem, when calculating the probability of slope instability, the fixed function is X is an independent standard normal random vector, F s (X) is the slope safety factor obtained based on the input variables. s When it is less than 0, the slope is in an unstable state, otherwise it is in a stable state.

[0083] The probability of slope failure p Fs It can be calculated by formula (2).

[0084] p Fs =p[G(x)<0]=∫ G(x)＜0 Γ(·)f(x)dx (2)

[0085] Among them, p Fs is the probability value of slope instability; f(x) is the joint probability density function between variables related to the probability of slope instability; Γ(·) is the index function, which is 1 when G(X)<0 (slope safety factor is less than 1), otherwise it is 0; G(X)=0 represents the limit state surface, that is, the interface between the safety domain and the instability domain.

[0086] The specific calculation requires the use of Python and Geo-studio or proxy models to obtain data for calculation. For a sufficient number of input variable samples, formula (2) can be converted to formula (3).

[0087]

[0088] Among them, N s is the total number of all samples; X (j) is the jth sample.

[0089] Therefore, according to the slope safety factor corresponding to each sample, the slope instability probability of the unsaturated soil slope is determined, including: counting the number of slope safety factors less than the preset safety threshold among all slope safety factors; and determining the ratio of the number to the total number of all samples as the slope instability probability of the unsaturated soil slope. That is, as shown in formula (3).

[0090] In an exemplary embodiment, the present invention also provides a reliability analysis method for unsaturated soil slope based on GeoStudio intelligent proxy model, such as Figure 2 As shown, this embodiment includes the following steps:

[0091] S201, complete the measurement of sparse data points on the slope and the collection of existing data on similar sites.

[0092] S202, based on the sparse data points of the slope and the existing data of similar sites, obtain the parameter posterior distribution samples N corresponding to different SWCCs s .

[0093] S203, based on the geometric shape of the slope and soil layer parameters, the original analysis model of slope stability is established in Geo-studio.

[0094] S204, exporting the original slope stability analysis model as a modifiable “#.xml” text file.

[0095] S205, based on Python, replace the SWCC parameters in the “#.xml” text file with the i-th posterior sample (α i , m i , n i ), get the corresponding file "i-th.xml", set the initial value i = 1, until i = N t .

[0096] S206, get N t Slope stability analysis model corresponding to SWCC.

[0097] S207, using the AutoIt library in Python to control Geo-studio to run the "i-th.xml" files in sequence.

[0098] S208, extract all "i-th.xml" files and run to get the slope safety factor F t,i .

[0099] S209, based on the posterior samples corresponding to SWCC and the corresponding slope safety factor, the response surface and machine learning random forest algorithm proxy models are constructed.

[0100] Among them, if the sample size in S202 is N scan be obtained from N s Get N samples from t The slope stability analysis model is constructed with samples, so that N t The slope safety factor corresponding to the samples is calculated according to N t The proxy model is constructed by using samples and the corresponding slope safety factor, and then the remaining samples (N s -N t ) of the slope safety factor.

[0101] S210, calculate the slope instability probability p when considering SWCC variation Fs .

[0102] The whole process of the present invention is implemented as follows Figure 2 As shown in Figure 3, the influence of different SWCCs on slope reliability analysis is studied. The whole process mainly involves the transmission of uncertainty, that is, the uncertainty of SWCC is generated by statistical uncertainty and model uncertainty, which is further transmitted to the uncertainty of slope stability safety factor.

[0103] Among them, steps S203-S208 are the process of realizing automated slope reliability analysis by developing Python and Geo-studio to automatically run batch processing linkage software and using Python scripts, such as Figure 3 As shown, Figure 3 Schematic diagram of slope reliability analysis automated with Python scripts.

[0104] In a specific embodiment, based on the technical solution of the present invention, this embodiment selects a loess tableland slope in the southern part of a city to study the influence of SWCC on the reliability analysis of unsaturated soil slopes. The slope soil is Pleistocene Q2-Holocene Q4 soil, and all are silt loess. The slope elevation is 65m, the slope angle is 45.4°, the groundwater level is -80m, the horizontal extension distance of the slope modeling is 150m, and the vertical extension distance is 100m. The specific soil attribute parameters are shown in Table 1 below, where the data are all taken from the mean value within the soil layer range. Establish the following in the SLOPE / W module: Figure 4 The slope analysis model is shown.

[0105] Table 1

[0106]

[0107] The distribution of a large number of SWCC parameter samples obtained based on hierarchical Bayes is as follows Figure 5 As shown, Figure 5 This is a large number of slope SWCC parameter data maps obtained based on hierarchical Bayesian. Based on the above technical solution, calculations and construction of the proxy model were performed, and the following results were obtained by verification using test data. Figure 6The validation graph and the measurement index values ​​shown in the figure show that the accuracy of the proxy model construction is very high. Figure 6 This is a proxy model data verification diagram based on the calculation results of Python and Geo-studio. Figure 6 The figure (a) in the figure represents the polynomial surrogate model. Figure 6 Figure (b) shows the RF proxy model (machine learning proxy model) constructed by the machine learning random forest (RF) algorithm, in which the root mean square error (RMSE), mean absolute percentage error (MAPE) and determination coefficient (R-squared, R 2 ) to verify the effectiveness of the polynomial surrogate model and the RF surrogate model. In the polynomial surrogate model, RMSE = 0.0011, MAPE = 0.0692, R 2 =0.9927, in the RF proxy model: RMSE = 0.0010, MAPE = 0.0615, R 2 =0.9983. Based on the calculation results, we can further obtain Figure 7 The slope safety factor histogram shown in Figure 1 shows that the corresponding slope instability probability is calculated to be 3.4×10-3. Figure 7 This is a slope instability probability map calculated based on Python, Geo-stduio linkage and proxy model.

[0108] In order to overcome the shortcomings of existing slope reliability analysis, the present invention constructs an automated batch implementation method to significantly improve the accuracy and efficiency of large-scale parameter posterior sample combination analysis. The technical solution adopted by the present invention is mainly divided into four steps:

[0109] (1) Using the hierarchical Bayesian approach, a large number of posterior samples of SWCC parameters corresponding to the slope reliability analysis (extremely sparse observation data / slope measurement data for unsaturated soil slopes) are obtained to characterize the variability of SWCC.

[0110] (2) The SLOPE / W module in Geo-studio was used to complete the slope reliability analysis modeling, and the Python and Geo-studio linkage software was further developed to complete the batch processing and calculation of slope safety factors corresponding to a large number of SWCCs.

[0111] (3) Based on the results obtained by the linkage software of Python and Geo-studio, a proxy model for the rapid calculation of slope safety factors for a large number of SWCCs is constructed, thereby further improving the computational efficiency of slope reliability analysis.

[0112] (4) Based on (2) and (3), the probability of slope instability is calculated to achieve slope reliability analysis.

[0113] In the present invention, a multi-software linkage calculation framework is proposed: Python and Geo-Studio are deeply combined to form linkage software that can perform batch calculations. A fast calculation proxy model is proposed: by training and verifying accurate proxy models, the calculation time is shortened while maintaining analysis accuracy. Automated tool chain: independently develop full-process automated processing tools from input to result extraction to avoid human operation errors. Instability probability calculation: combining a large number of SWCCs and probability analysis methods, considering the impact of different SWCCs on slope reliability analysis, and conducting rapid quantitative evaluation of slope reliability, which is widely applicable to slope design, stability assessment, risk prediction and engineering optimization and other fields.

[0114] Therefore, compared with the prior art, the present invention:

[0115] (1) High degree of automation: Traditional technologies rely on manual generation of random parameters and step-by-step analysis, which is cumbersome and error-prone. However, the present invention uses Python and Geo-Studio’s linked calculation software to achieve full-process automated analysis of SWCC parameter uncertainty, including parameter generation, calculation execution, and result extraction.

[0116] (2) Efficient batch computing capability: The present invention supports batch analysis of a large number of SWCC parameter posterior samples, which significantly improves the computing efficiency, while traditional methods cannot respond quickly in large-scale parameter combination analysis.

[0117] (3) Improved accuracy and reliability: The present invention generates posterior samples through the Bayesian method, comprehensively quantifies the impact of the uncertainty of SWCC parameters on slope stability, and can more accurately reflect the diversity of actual slope conditions. Combined with the probability analysis method, the probability of slope instability is quantitatively calculated, providing a scientific basis for reliability analysis and avoiding the shortcomings of the existing technology that relies on a single parameter.

[0118] (4) Innovative technology integration: For the first time, Python and Geo-Studio are deeply linked to form a highly integrated computing framework, breaking through the problem of insufficient coordination between software in existing technologies. An efficient proxy model is built and combined with automation tools to achieve full-process rapid calculation of slope reliability analysis considering SWCC uncertainty.

[0119] When applying the reliability analysis method of unsaturated soil slope based on GeoStudio intelligent proxy model provided by the present invention, it is not necessary to Figure 1 The steps are executed in the order shown. The specific execution order of the steps can be determined according to needs, and the present invention does not limit this.

[0120] The above is a reliability analysis method for an unsaturated soil slope based on a GeoStudio intelligent proxy model provided by one or more embodiments of the present invention. Based on the same idea, the present invention also provides a corresponding reliability analysis device for an unsaturated soil slope based on a GeoStudio intelligent proxy model, such as Fig. 9 shown.

[0121] Figure 8 A schematic diagram of a reliability analysis device for an unsaturated soil slope based on a GeoStudio intelligent proxy model provided by the present invention, the device 800 comprises:

[0122] The determination module 801 is used to combine the slope measurement data of the unsaturated soil slope with the existing data to determine multiple samples of soil-water characteristic curve model parameters; the existing data are measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope.

[0123] The export module 802 is used to construct an original analysis model of slope stability on GeoStudio software according to the slope measurement data and the soil-water characteristic curve model, and export the original analysis model of slope stability as a text file.

[0124] The linkage module 803 is used to divide the multiple samples into two groups, control the GeoStudio software to automatically run the text file through the programming language script, and batch calculate the slope safety factor corresponding to the first group of samples.

[0125] The construction module 804 is used to construct a proxy model through the first group of samples and the corresponding slope safety factors, and calculate the slope safety factors corresponding to the second group of samples through the proxy model.

[0126] The calculation module 805 is used to determine the slope instability probability of the unsaturated soil slope according to the slope safety factors corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

[0127] For the specific definition of the reliability analysis device for unsaturated soil slopes based on the GeoStudio intelligent proxy model, please refer to the definition of the reliability analysis method for unsaturated soil slopes based on the GeoStudio intelligent proxy model in the above text, which will not be repeated here. Each module in the above-mentioned reliability analysis device for unsaturated soil slopes based on the GeoStudio intelligent proxy model can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0128] The present invention also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 A reliability analysis method for unsaturated soil slopes based on the GeoStudio intelligent proxy model is provided.

[0129] The present invention also provides Fig. 9 The structural diagram of the computer device shown in FIG. Fig. 9 As shown in the figure, at the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 A reliability analysis method for unsaturated soil slopes based on the GeoStudio intelligent proxy model is provided.

[0130] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0131] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present invention.

Claims

1. A reliability analysis method for unsaturated soil slope based on GeoStudio intelligent proxy model, characterized in that: include: Combining the slope measurement data of the unsaturated soil slope with existing data to determine multiple samples of soil-water characteristic curve model parameters; the existing data are measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope; According to the slope measurement data and the soil-water characteristic curve model, an original analysis model of slope stability is constructed on GeoStudio software, and the original analysis model of slope stability is exported as a modifiable text file; Divide the multiple samples into two groups, control the GeoStudio software to automatically run the text file through a programming language script, and batch calculate the slope safety factor corresponding to the first group of samples; A proxy model is constructed by using the first group of samples and the corresponding slope safety factors, and the slope safety factors corresponding to the second group of samples are respectively calculated by using the proxy model; The slope instability probability of the unsaturated soil slope is determined according to the slope safety factors corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

2. The method according to claim 1, characterized in that The method of controlling the GeoStudio software to automatically run the text file through a programming language script and batch calculating the slope safety factor corresponding to the first group of samples includes: Assigning the first group of samples to the model parameters of the soil-water characteristic curve model in the text file through a programming language script to obtain multiple model files; The GeoStudio software is controlled by the software library corresponding to the programming language to run each model file in sequence to obtain the slope safety factor corresponding to the first group of samples.

3. The method according to claim 2, characterized in that The software library corresponding to the programming language controls the GeoStudio software to run each model file in sequence to obtain the slope safety factor corresponding to the first group of samples, including: For any sample in the first group of samples, the GeoStudio software is controlled by a software library corresponding to the programming language to run the model file corresponding to the sample to obtain the safety factors of multiple sliding surfaces; The smallest safety factor among the safety factors of the multiple sliding surfaces is determined as the slope safety factor corresponding to the sample.

4. The method according to claim 1, characterized in that: The method of constructing a proxy model using the first group of samples and the corresponding slope safety factor includes: Dividing the first group of samples and the corresponding slope safety factors to obtain a training data set and a test data set; The proxy model is constructed using the training data set, and the accuracy of the proxy model is verified using the test data set.

5. The method according to claim 4, characterized in that The proxy model includes a polynomial proxy model; and constructing the proxy model through the training data set includes: Fitting a second-order polynomial without cross terms to the training data set to determine coefficients of the second-order polynomial without cross terms; The fitted second-order polynomial without cross terms is determined as the polynomial surrogate model.

6. The method according to claim 4, characterized in that The proxy model includes a machine learning proxy model; and constructing the proxy model through the training data set includes: Based on the training data set, the machine learning agent model is constructed by a machine learning random forest algorithm.

7. The method according to claim 1, characterized in that Determining the slope instability probability of the unsaturated soil slope according to the slope safety factor corresponding to all samples includes: Count the number of slope safety factors less than a preset safety threshold among all slope safety factors; The ratio of the number to the total number of all samples is determined as the slope instability probability of the unsaturated soil slope.

8. A reliability analysis device for unsaturated soil slope based on GeoStudio intelligent proxy model, characterized in that: include: A determination module is used to combine the slope measurement data of the unsaturated soil slope with existing data to determine multiple samples of soil-water characteristic curve model parameters; the existing data is measurement data of sites with similar geological history and depositional environment and the same soil type as the unsaturated soil slope; An export module is used to construct an original analysis model of slope stability on GeoStudio software according to the slope measurement data and the soil-water characteristic curve model, and export the original analysis model of slope stability as a modifiable text file; A linkage module is used to divide the multiple samples into two groups, control the GeoStudio software to automatically run the text file through a programming language script, and batch calculate the slope safety factor corresponding to the first group of samples; A construction module, used to construct a proxy model through the first group of samples and the corresponding slope safety factors, and calculate the slope safety factors corresponding to the second group of samples through the proxy model; The calculation module is used to determine the slope instability probability of the unsaturated soil slope according to the slope safety factors corresponding to all samples; the slope instability probability is used to characterize the reliability of the unsaturated soil slope.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images