Method and apparatus for assessing highway resilience under rainfall landslide conditions

Abstract

This invention relates to a method and equipment for probabilistic assessment of highway resilience under rainfall-induced landslide conditions, comprising the following steps: S1, calibrating a random rainfall model and the probability density distribution function of soil and rock parameters; S2, numerically simulating the sliding distance of the slope using a two-stage analysis of the finite element method and the material point method, and constructing a sample library of highway residual functions; S3, constructing a two-stage predictive proxy model for highway residual functions; S4, randomly sampling soil and rock parameters and rainfall events, and predicting highway residual functions based on the two-stage predictive proxy model; S5, constructing a probabilistic ladder recovery model for highway functions based on a decision tree, randomly sampling the recovery patterns of the highway after different landslide events, and calculating the highway resilience index based on the predicted value of the highway residual functions; S6, repeating step S5 multiple times to solve for the mean and standard deviation of the highway resilience index. Compared with existing technologies, this invention can achieve rapid and accurate assessment of the resilience level of slope highways.

Description

Technical Field

[0001] This invention belongs to the field of landslide disaster prevention and control technology, and in particular relates to a method and equipment for assessing the resilience probability of highways under rainfall-induced landslide conditions. Background Technology

[0002] Transportation system resilience has been incorporated into the safety construction evaluation index system. Mastering the assessment methods for transportation infrastructure resilience and breaking through key technologies such as improving the resilience of transportation infrastructure are the main tasks for enhancing the high-quality construction and maintenance technology level of infrastructure. The literature "Research on the Resilience of Transportation Infrastructure: Origin, Development and Future" (Liu Lanjian and Dong Yongqiang, China Safety Science Journal, 2023, Vol. 33(11):45-51) points out that enhancing the resilience of transportation infrastructure, reducing the adverse effects of shock events and quickly restoring it to an ideal state are directly related to the stability and sustainability of transportation. Therefore, rapid assessment of the resilience of slope highway systems is of great significance to the development of transportation system safety construction and improving the resilience design level of my country's transportation infrastructure.

[0003] Currently, a large number of studies have been conducted on evaluating engineering projects using resilience theory. The literature "Current Status and Prospect of Road Traffic Infrastructure Resilience Research" (Huang Xiaoming and Zhao Rundong, Journal of Jilin University (Engineering Science Edition), 2022, 53(06): 1529-1549) points out that most existing traffic resilience studies focus on the capacity of the macro road network and even the transportation network level, neglecting the resilience of the structures such as highways in the traffic infrastructure. The difficulty lies in the low degree of quantification of the residual function assessment of structures such as highways. The literature "Vulnerability Assessment Model of Disaster-Bearing Bodies and Landslide Disaster Risk Index" (Wu Yue, Liu Dongsheng, Lu Xin, et al., Rock and Soil Mechanics, 2011, Vol. 32(08): 2487-2492+2499) points out that the quantitative assessment of the functional damage degree of disaster-bearing bodies is a bottleneck problem restricting the risk assessment of landslide disasters, and its quantification degree determines the accuracy and scientific nature of the resilience assessment of highway infrastructure. Furthermore, highways adjacent to slopes are susceptible to the impact of events such as rainfall. Highway slope resilience assessment needs to consider the uncertainties of actual rainfall events, soil and rock parameters, and recovery patterns. Combining resilience theory with probabilistic analysis can provide a more comprehensive assessment of slope resilience. After calculating the slope resilience index through a resilience model, the Monte Carlo simulation-based probabilistic resilience assessment method can calculate the uncertainty range of the slope resilience index, providing richer information for engineering decisions. Although existing resilience models can comprehensively consider uncertainties in the system, they mostly treat residual functions as random variables rather than physical-based numerical simulations, which is not practically applicable to slope design. In summary, due to complex uncertainties and low quantification of residual functions, existing highway slope resilience assessments still suffer from incomplete theoretical frameworks and low computational efficiency. Therefore, it is necessary to propose a new assessment method to accurately consider the functional losses of highway facilities and quickly achieve highway slope resilience assessment. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a method and equipment for probabilistic assessment of highway resilience under rainfall and landslide conditions, further improving the theoretical basis of highway infrastructure resilience assessment and increasing computational efficiency.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] This invention provides a method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions, comprising the following steps:

[0007] S1. Collect historical rainfall data, and statistically analyze the mean and coefficient of variation of soil and rock parameters, and calibrate the random rainfall model and the probability density distribution function of soil and rock parameters;

[0008] S2. Based on the random rainfall model and probability density distribution function of soil and rock parameters calibrated in step S1, soil and rock parameters and rainfall events are uniformly extracted, and the sliding distance of the slope is numerically simulated using a two-stage analysis of finite element method-material point method to construct a sample library of highway residual functions.

[0009] S3. Based on the highway residual function sample library, a two-stage prediction proxy model for highway residual function is constructed through a feedforward neural network to determine whether the highway function is affected by the slope and the degree of slope influence under different parameter conditions.

[0010] S4. Randomly sample soil and rock parameters and rainfall events, and predict highway residual function based on the two-stage prediction proxy model of the highway residual function;

[0011] S5. Construct a probabilistic ladder recovery model for highway function based on decision tree, randomly select the recovery mode of highway after different landslide events, and calculate the highway resilience index based on the predicted value of highway residual function obtained in step S4.

[0012] S6. Repeat step S5 multiple times to calculate the mean and standard deviation of the highway resilience index and obtain the assessment results of the resilience level of highway infrastructure.

[0013] Further, in step S1, each rainfall data point includes rainfall intensity and duration. The univariate marginal distributions of rainfall intensity and duration are fitted using a generalized Pareto distribution. Then, the Frank Copula function is used to connect the univariate marginal distributions to obtain a binary joint distribution. The random rainfall model is then calibrated using the maximum likelihood method.

[0014] Furthermore, in step S1, the probability density distribution function of the soil and rock parameters is calibrated using a log-normal distribution.

[0015] Furthermore, in step S2, the construction process of the highway residual function sample library is as follows:

[0016] S201. Use the finite element method to perform seepage and stability analysis on the slope, search for the critical failure point where the safety factor is initially less than 1, and derive the corresponding seepage field and stress field.

[0017] S202. The material point method is used to analyze the motion and deformation of the slope after failure. The seepage field and stress field derived in step S201 are imported as initial conditions, and the slope sliding distance is solved by the control equation.

[0018] S203. Calculate the number of blocked highway lanes based on the slope sliding distance, and then solve for the highway residual function to construct the highway residual function sample library.

[0019] Furthermore, in step S203, the calculation method for the residual function of the highway is as follows:

[0020]

[0021] Among them, Q r For the residual function of the highway, n d The number of lanes blocked by the landslide is m, and the total number of lanes on the highway is m.

[0022] Further, in step S3, samples in the highway residual function sample library with a highway residual function value less than 1 are marked as failed samples. The specific construction process of the two-stage prediction proxy model for highway residual function is as follows:

[0023] Based on all samples in the highway residual function sample library, a binary classification surrogate prediction model is established using a feedforward neural network to determine whether the highway function is affected by the slope under different parameter conditions.

[0024] Based on the failure samples in the highway residual function sample library, a multi-class surrogate prediction model is established using a feedforward neural network to determine the degree to which highway function is affected by slope under different parameter conditions.

[0025] Furthermore, in step S4, random sampling of soil and rock parameters and rainfall events is performed using Monte Carlo simulation.

[0026] Furthermore, in step S5, the probabilistic ladder recovery model for highway function is as follows:

[0027]

[0028]

[0029]

[0030]

[0031] Where Q(t) represents the function in the recovery process, i represents the stage number, and ΔQ i Let Δt be the functional increment for the i-th stage, t0 be the initial repair time, and Δt be the function increment for the i-th stage. i Let t1 be the time required for the repair in the i-th stage, t1 be the time when the repair is completed, f(a) be the probability mass function of recovery mode A, and Dir(H|α=1) be the Pindyrick distribution, where H={h i , iid|i=1,...,K A} represents the percentage of recovery time, K A h represents the total number of stages in recovery mode A. i Let n be the percentage of recovery time for the i-th stage. dLet Γ be the number of lanes blocked by the landslide, Γ be the Gamma function, and α be the concentration parameter.

[0032] Furthermore, in step S5, the formula for calculating the highway resilience index is as follows:

[0033]

[0034] Where R is the highway resilience index, Q r For the residual function of the highway, ΔQ i For the functional increment of the i-th stage, h i+1 Let K be the percentage of recovery time in the (i+1)th stage, and K be the total number of stages.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a program stored in the memory, wherein the processor executes the program to implement the above-described method.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. This invention employs a two-stage analysis using the finite element method and the material point method to numerically simulate the sliding distance of slopes, constructing a highway residual function sample library. Based on this sample library, a two-stage predictive surrogate model for highway residual function is built using a feedforward neural network. This model is used to sequentially determine whether and to what extent highway function is affected by the slope under different parameter conditions. This improves the quantification of highway residual function analysis in slope highway toughness assessment, avoids the cumbersome mechanical calculations that may be involved in predicting landslide events, and improves the solution efficiency for predicting highway slope failure events under rainfall conditions. By combining numerical simulation with the surrogate model, the toughness level of highway slopes can be quickly and accurately assessed, helping engineering managers to provide more comprehensive decision-making and guidance for the toughness design of highway slopes.

[0038] 2. This invention considers the functional characteristics of highway infrastructure and constructs a probabilistic ladder recovery model for highway function based on decision trees. It randomly selects the recovery mode of the highway after different landslide events and calculates the highway resilience index based on the predicted value of the highway residual function. The traditional resilience index is transformed into the sum of the proportion of residual function and the functional increment during the recovery period. The predicted value of the highway residual function is obtained by using Monte Carlo simulation of randomly sampled geotechnical parameters and rainfall events, based on a two-stage predictive proxy model of highway residual function. The above method can calculate the uncertainty range of slope resilience index, providing richer information for engineering decision-making and further improving the theoretical framework for slope highway resilience assessment. Attached Figure Description

[0039] Figure 1 This is a flowchart of the method of the present invention;

[0040] Figure 2 This is an example diagram of pore pressure distribution at critical failure of a numerical sample.

[0041] Figure 3 This is an example of a displacement distribution cloud map for a numerical sample.

[0042] Figure 4 Histogram for predicting the distribution of residual functions of highways;

[0043] Figure 5 n is the number of lanes blocked by the landslide. d The figures show the highway function recovery curves under four recovery modes at time 3, with each recovery mode including three simulation examples.

[0044] Among them, (5a) is a single-stage restoration example of highway function (△Q1=75%), (5b) is a two-stage restoration example of highway function (△Q1=25%, △Q2=50%), (5c) is a two-stage restoration example of highway function (△Q1=50%, △Q2=25%), and (5d) is a three-stage restoration example of highway function (△Q1=△Q2=△Q3=25%).

[0045] Figure 6 Let be the histogram of the distribution of highway resilience index, where E(R) is the mean of highway resilience index R, and σ(R) is the standard deviation of highway resilience index R. Detailed Implementation

[0046] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0047] Example:

[0048] This embodiment provides a method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions. It is based on a typical mean slope example from the literature, assuming a slope height of 10m, a slope angle of 35.5°, and sandy soil, with a 15-meter-wide four-lane highway adjacent to the slope. The specific process is as follows: Figure 1 As shown, it includes the following steps:

[0049] S1. Collect historical rainfall data, and statistically analyze the mean and coefficient of variation of soil and rock parameters to calibrate the random rainfall model and the probability density distribution function of soil and rock parameters.

[0050] The random rainfall model was constructed based on hourly rainfall data and research findings from the St. James Complex in Singapore from 1980 to 2010, using a 9-hour event interval and a rainfall intensity threshold of 5 mm / h.

[0051] The intensity and duration frequency distribution characteristics of independent rainfall events were statistically analyzed. The obtained rainfall intensity was mainly distributed between 3-63 mm / h, and the rainfall duration was mainly distributed between 1-21 h. Based on the data distribution characteristics, this embodiment uses a generalized Pareto distribution to fit the univariate marginal distribution of rainfall intensity q and duration t. The generalized Pareto distribution formula can be written in the following form:

[0052]

[0053] In the formula, k, μ and σ represent shape, position and scale parameters, respectively, and r = q or t.

[0054] Then, the Frank Copula function is used to connect the univariate marginal distributions to obtain the bivariate joint distribution, as shown in the following expression:

[0055]

[0056] In the formula, δ is the Frank Copula parameter, and r = {q, t}.

[0057] The rainfall distribution parameters obtained by the maximum likelihood method are shown in Table 1.

[0058] Table 1. Binary Joint Distribution Parameters of Rainfall Events

[0059]

[0060] By treating soil and rock parameters as random variables, the mean and coefficient of variation of these parameters are statistically analyzed, and the probability density distribution function of the soil and rock parameters is calibrated. The uncertain soil parameters considered include: effective internal friction angle. Van Genuchten model parameters α, n, and saturated permeability coefficient k s Based on typical sandy soil values, the mean and coefficient of variation of various uncertain soil parameters are shown in Table 2.

[0061] Table 2. Mean, coefficient of variation, and distribution of random variables for soil parameters.

[0062]

[0063] S2. Uniformly extract soil and rock parameters and rainfall events, and use the finite element method-material point method two-stage analysis to numerically simulate the sliding distance of the slope and construct a highway residual function sample library.

[0064] To ensure the applicability of the sample database, the range of values ​​for the random variables of soil parameters is [x]. u -4x σ x u +4x σ ], where x u and x σThese represent the mean and standard deviation of a random variable within a standard normal space. The sample values ​​for rainfall intensity and duration range from [0, 72 mm / h] to [0, 48 h]. A uniform sample of 1000 values ​​is drawn from these ranges to generate 500 six-dimensional input variables containing soil parameters and rainfall conditions.

[0065] Combining the finite element method with the material point method to consider the impact of slope failure on highway function under rainfall conditions can improve the quantification of residual function analysis in highway slope toughness assessment. The construction process of the highway residual function sample library is as follows:

[0066] S201. For a given input variable, the finite element method is used to perform seepage and stability analysis on the slope. The initial safety factor less than 1 is used as the critical failure point, and the corresponding seepage field and stress field are derived. An example diagram of pore pressure distribution at critical failure of a numerical sample is shown below. Figure 2 As shown;

[0067] S202. The material point method is used to analyze the motion and deformation of the slope after failure. The seepage field and stress field derived in step S201 are imported as initial conditions. The slope sliding distance is solved by the governing equations such as mass and momentum conservation. The displacement distribution cloud map of a certain numerical sample is shown below. Figure 3 As shown;

[0068] S203. Calculate the number of blocked highway lanes based on the calculated slope sliding distance, then solve for the highway residual function, and construct a highway residual function sample library. The specific calculation method for highway residual function is as follows:

[0069]

[0070] Among them, Q r For the residual function of the highway, n d The number of lanes blocked by the landslide is m, and the total number of lanes on the highway is m.

[0071] The slope sliding distance of this sample was 2.74m, and the number of lanes blocked by the landslide was n. d =1, Highway residual function Q r =75%. The residual function is used as the output variable and combined with the input variables to form a sample. The above operation is repeated for 500 input variables to form a highway residual function sample library.

[0072] S3. Based on the highway residual function sample library, a two-stage prediction proxy model for highway residual function is constructed through a feedforward neural network.

[0073] Samples with a highway residual function value less than 1 in the highway residual function sample library are marked as failed samples. The specific construction process of the two-stage prediction surrogate model for highway residual function is as follows:

[0074] S301. Based on all samples in the highway residual function sample library, a binary classification surrogate prediction model is established using a feedforward neural network to determine whether highway function is affected by slope under different parameter conditions. The first stage is to divide all samples in the highway residual function sample library into two categories according to whether the residual function is less than 1. The classified training samples are then divided into a test sample set and a training sample set in a 2:8 ratio to construct a binary classification surrogate prediction model based on a feedforward neural network.

[0075] S302. Based on failure samples in the highway residual function sample library, a multi-classification surrogate prediction model is established using a feedforward neural network to determine the degree to which highway function is affected by slope under different parameter conditions. Sensitivity analysis shows that the number of lanes n blocked by landslides... d The maximum value is 3, so in the second stage, all samples with residual function less than 1 in the sample library are divided into three classes. The classified training samples are then divided into a test sample set and a training sample set in a 2:8 ratio to construct a multi-class surrogate prediction model based on a feedforward neural network.

[0076] A two-stage classification surrogate prediction model based on a feedforward neural network, built using numerical samples, can avoid the tedious mechanical calculations involved in predicting landslide events. The trained random forest surrogate model can be evaluated for its classification accuracy based on the confusion matrix. The two-stage classification surrogate prediction model determines whether landslides will occur under different rainfall and soil conditions, and the residual functional magnitude of the highway after landslide damage, thus improving the solution efficiency for predicting highway slope failure events under rainfall conditions. In this embodiment, the AUC value for the first stage calculated based on the confusion matrix of the test sample set is 0.99, and the average AUC value for the second stage is 0.91, indicating that the two-stage prediction surrogate model has high prediction accuracy.

[0077] S4. Randomly sampled soil and rock parameters and rainfall events are used to predict highway residual function based on a two-stage prediction proxy model.

[0078] Through Monte Carlo simulation, a set of soil parameters θ is sampled based on the probability density distribution function of the soil parameters, and a set of rainfall events r during the service period is sampled based on the random rainfall model. d(θ,r) is the slope sliding distance when the rainfall condition is r and the soil parameters θ are given, the single lane width is b, and S[d(θ,r)] is the indicator function of the residual function of the highway, which is obtained through the landslide classification proxy prediction model and is defined as follows:

[0079]

[0080] The input vector contains soil parameters and rainfall parameters. The highway residual function is predicted using the indicator function S[d(θ,r)]. The highway residual function prediction distribution histogram is shown below. Figure 4 As shown.

[0081] S5. Construct a probabilistic ladder recovery model for highway function based on decision tree, randomly select the recovery mode of highway after different landslide events, and calculate the highway resilience index based on the predicted value of highway residual function obtained in step S4.

[0082] Considering the functional characteristics of highways, a step function is used to describe their function during the recovery process:

[0083]

[0084] In the formula, Q(t) represents the function in the recovery process, i represents the stage number, and ΔQ i Let Δt be the functional increment for the i-th stage, t0 be the initial repair time, and Δt be the function increment for the i-th stage. i Let t1 be the time required for the repair in the i-th stage, and t1 be the time when the repair is completed.

[0085] The step shape of the step function describes the recovery patterns of the highway. When the maximum number of blocked lanes is n, the total number of recovery patterns is 2. n Given the number of congested lanes n d The probability mass function f(a) for recovery mode A is described as follows:

[0086]

[0087] The recovery time is normalized, and the recovery time proportion of each stage is described based on a uniform distribution as follows:

[0088]

[0089]

[0090] In the formula, Dir(H|α=1) is the Pindyrickley distribution, and H={h i , iid|i=1,...,K A} represents the percentage of recovery time, K A h represents the total number of stages in recovery mode A. i Let Γ be the proportion of recovery time in the i-th stage, Γ be the Gamma function, and α be the concentration parameter.

[0091] Q(t) is a deterministic expression for the evolution of highway function, and is related to f(a), Dir(H|α=1) and Together they constitute a probabilistic ladder recovery model.

[0092] The number of lanes n blocked by the landslide d The maximum value is 3. This step recovery model has a total of 8 recovery modes, as shown in Table 3.

[0093] Table 3 Step parameters of the step restoration model

[0094]

[0095]

[0096] The number of lanes n blocked by the landslide d When the value is 3, the highway function recovery curves under the four recovery modes are as follows: Figure 5 As shown, each recovery mode includes three simulation examples: Simulation 1, Simulation 2, and Simulation 3. It can be observed that as the residual function decreases, the shape of the recovery curve becomes more diverse and uncertain, resulting in changes in highway toughness under different recovery mode selections.

[0097] By using a step function, the traditional resilience index is transformed into the sum of the proportion of residual function and the functional increment during the recovery period. This clearly indicates the degree of highway resistance to slope disasters and the contribution of human restoration processes to highway resilience. It is found that accurate characterization of highway residual function based on physical processes is crucial for the quantitative assessment of highway resilience. The formula for calculating the highway resilience index R is as follows:

[0098]

[0099] The Monte Carlo simulation-based probabilistic resilience assessment method can calculate the uncertainty range of slope resilience indices, providing richer information for the resilience design of highway slopes and further improving the theoretical framework for highway slope resilience assessment. When the number of blocked lanes is equal to 1, 2, and 3, the probability mass functions of the corresponding recovery modes are 1, 0.5, and 0.25, respectively. For different residual function conditions, based on the recovery modes and the proportion of recovery time at each stage randomly sampled from the probabilistic ladder recovery model, the slope resilience index is calculated according to the residual function, the functional increment during the recovery period, and the proportion of recovery time at each stage.

[0100] S6. Repeat step S5 multiple times to calculate the mean and standard deviation of the highway resilience index and obtain the assessment results of the resilience level of highway infrastructure.

[0101] The mean E(R) and standard deviation σ(R) of the slope resilience index are obtained based on the following formulas:

[0102]

[0103]

[0104] In the formula, R iLet N represent the i-th group of soil and rock parameters and the slope toughness index under the rainfall event, and let N represent the number of times step S5 is repeated. In this embodiment, repeating the step 1000 times will yield 1000 groups of soil and rock parameters and the slope toughness index under the rainfall event. The slope toughness index distribution histogram is shown below. Figure 6 As shown, the mean resilience index E(R) of the slope is 0.8782, and the standard deviation of the resilience index σ(R) is 0.2107. This means that during the restoration process, the road's usability is 87%, which means that approximately 3.5 lanes can operate normally on average.

[0105] If the above methods are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0106] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions, characterized in that, Includes the following steps: S1. Collect historical rainfall data, and statistically analyze the mean and coefficient of variation of soil and rock parameters, and calibrate the random rainfall model and the probability density distribution function of soil and rock parameters; S2. Based on the random rainfall model and probability density distribution function of soil and rock parameters calibrated in step S1, soil and rock parameters and rainfall events are uniformly extracted, and the sliding distance of the slope is numerically simulated using a two-stage analysis of finite element method-material point method to construct a sample library of highway residual functions. S3. Based on the highway residual function sample library, a two-stage prediction proxy model for highway residual function is constructed through a feedforward neural network to determine whether the highway function is affected by the slope and the degree of slope influence under different parameter conditions. S4. Randomly sample soil and rock parameters and rainfall events, and predict highway residual function based on the two-stage prediction proxy model of the highway residual function; S5. Construct a probabilistic ladder recovery model for highway function based on decision tree, randomly select the recovery mode of highway after different landslide events, and calculate the highway resilience index based on the predicted value of highway residual function obtained in step S4. S6. Repeat step S5 multiple times to calculate the mean and standard deviation of the highway resilience index and obtain the assessment results of the resilience level of highway infrastructure. In step S2, the construction process of the highway residual function sample library is as follows: S201. Use the finite element method to perform seepage and stability analysis on the slope, search for the critical failure point where the safety factor is initially less than 1, and derive the corresponding seepage field and stress field. S202. The material point method is used to analyze the motion and deformation of the slope after failure. The seepage field and stress field derived in step S201 are imported as initial conditions, and the slope sliding distance is solved by the control equation. S203. Calculate the number of blocked highway lanes based on the slope sliding distance, then solve for the highway residual function and construct the highway residual function sample library; In step S203, the calculation method for the residual function of the highway is as follows: in, For the remaining functions of the highway, The number of lanes blocked by the landslide. This represents the total number of lanes on the highway. In step S3, samples in the highway residual function sample library with a highway residual function value less than 1 are marked as failed samples. The specific construction process of the two-stage prediction surrogate model for highway residual function is as follows: Based on all samples in the highway residual function sample library, a binary classification surrogate prediction model is established using a feedforward neural network to determine whether the highway function is affected by the slope under different parameter conditions. Based on the failure samples in the highway residual function sample library, a multi-class surrogate prediction model is established using a feedforward neural network to determine the degree to which highway function is affected by slope under different parameter conditions; in step S5, the probabilistic ladder recovery model of highway function is as follows: in, This indicates the function during the recovery process. Indicates the number of stages. For the first Functional increments at each stage To repair the initial moment, For the first The time required for each stage of repair To complete the repair, For recovery mode The probability mass function, It is distributed in Pindirich. As a percentage of recovery time, For recovery mode Total number of stages, For the first The percentage of recovery time in each stage The number of lanes blocked by the landslide. For the Gamma function, For concentration parameters; In step S5, the formula for calculating the highway resilience index is as follows: in, For highway resilience indicators, For the remaining functions of the highway, For the first Each stage of functional increments For the first +1 percentage of recovery time in each stage This represents the total number of stages.

2. The method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions according to claim 1, characterized in that, In step S1, each rainfall data point includes rainfall intensity and duration. The univariate marginal distributions of rainfall intensity and duration are fitted using a generalized Pareto distribution. Then, the Frank Copula function is used to connect the univariate marginal distributions to obtain a binary joint distribution. The random rainfall model is then calibrated using the maximum likelihood method.

3. The method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions according to claim 1, characterized in that, In step S1, the probability density distribution function of the soil and rock parameters is calibrated using a log-normal distribution.

4. The method for probabilistic assessment of highway resilience under rainfall-induced landslide conditions according to claim 1, characterized in that, In step S4, random sampling of soil and rock parameters and rainfall events is performed using Monte Carlo simulation.

5. An electronic device comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-4.

Images