Rock-soil composite landslide dynamic evaluation method and system based on multi-source data fusion

Abstract

The present application provides a rock-soil composite landslide dynamic evaluation method and system based on multi-source data fusion, relates to the field of slope stability evaluation and geological disaster prevention technology of geotechnical engineering, and sets up a collection point in the landslide area, obtains the shear strength parameters of each point in the natural and saturated states, and constructs a physical model; the parameters are input into the model to obtain the safety factor, and the parameters are divided into two groups of natural stability and saturated instability according to the comparison with the threshold value; the probability distribution of the two groups of parameters is determined respectively, Monte Carlo sampling is carried out, the new parameter group is input into the model to obtain the safety factor sample value, and the joint reliability evaluation index is obtained by statistics; at the same time, the sensitivity analysis is carried out to determine the weight, the weighted sum of the single reliability of each parameter is calculated, and the weighted reliability evaluation index is obtained; finally, the two indexes are compared with the precision threshold and the efficiency threshold, and the index meeting the requirements is selected as the final evaluation result.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering and geological disaster prevention technology, specifically to a dynamic assessment method and system for geotechnical composite landslides based on multi-source data fusion. Background Technology

[0002] Landslides are frequent and highly destructive geological hazards worldwide, and the accuracy of their reliability assessment is directly related to the effectiveness of risk prevention and control.

[0003] Traditional deterministic analysis methods, based on fixed parameter values, struggle to characterize the heterogeneity and uncertainty of soil and rock masses. To address this, researchers have introduced probabilistic analysis methods, treating soil and rock parameters as random variables for statistical analysis. In recent years, machine learning techniques have also been applied to landslide risk assessment. For example, patent document CN202111268460.9 utilizes models such as mixture Gaussian clustering and neural networks to evaluate and predict landslide hazard.

[0004] However, existing methods still have the following technical shortcomings: shear strength parameters evolve dynamically with slope conditions (stable / instable), while existing models mostly adopt static distribution assumptions, which cannot reflect parameter abrupt changes caused by external disturbances such as rainfall; they simply assume that parameters follow a normal or log-normal distribution, lacking a systematic identification of real distribution characteristics such as skewness and multimodality; they have not established an effective correlation with engineering sampling strategies, lacking differentiated sampling criteria for different hydrogeological conditions; they have failed to deeply analyze the complex correlation between cohesion and internal friction angle and their dynamic impact under different conditions; and they lack a systematic comparison and selection criteria for the applicability of different calculation methods.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic assessment method and system for rock and soil composite landslides based on multi-source data fusion, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The dynamic assessment method for soil-rock composite landslides based on multi-source data fusion includes the following steps: Several data collection points were set up in the target landslide area, and the shear strength parameters of each data collection point were obtained under both natural and saturated conditions. A physical model was constructed based on the topography and rock layer distribution characteristics of the landslide. Input the shear strength parameters of each sampling point under the two states into the physical model to obtain the safety factor of each sampling point under the two states. Determine the stable state based on the safety factor. Statistically count the shear strength parameters of the sampling points determined to be stable under the natural state as the natural stable state parameter group, and the shear strength parameters of the sampling points determined to be unstable under the saturated state as the saturated unstable state parameter group. For the parameter sets of natural stable state and saturated unstable state, the probability distribution type of each shear strength parameter in each set is determined. Monte Carlo random sampling is performed according to the probability distribution type. The generated new shear strength parameter set is input into the physical model to calculate the corresponding safety factor sample value. The joint reliability evaluation index under the two states is obtained based on the statistical results of all safety factor sample values. For the parameter groups of natural stable state and saturated unstable state, the weight coefficient of each parameter is determined according to the degree of influence of each shear strength parameter on landslide stability in each group, and the individual reliability evaluation index of each shear strength parameter is calculated. The weight is summed with the corresponding individual reliability evaluation index to obtain the weighted reliability evaluation index for the two states. The joint reliability evaluation index and the weighted reliability evaluation index for the two states are compared with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, the joint reliability evaluation index is selected as the final reliability evaluation index.

[0008] Furthermore, several sampling points were set up in the target landslide area. When obtaining the shear strength parameters of each sampling point under natural and saturated conditions, a typical profile consistent with the main sliding direction was selected in the target landslide area. The typical profile runs through the front, middle and rear edges of the landslide. Several sampling points were set up along the typical profile, and soil and rock samples were obtained at each sampling point under natural and saturated conditions.

[0009] Furthermore, the collected soil and rock samples were tested to obtain the shear strength parameters of each sampling point in both natural and saturated states. The shear strength parameters include unit weight, cohesion, and internal friction angle.

[0010] Furthermore, based on the topography and rock strata distribution characteristics of the target landslide, a physical model of the target landslide is constructed. The specific steps are as follows: Based on landslide topographic data, a two-dimensional limit equilibrium geometric model for landslide analysis was established. Based on rock layer distribution data, the landslide rock layer sequence was divided from top to bottom into an overlying sliding layer and an underlying non-sliding bedrock layer, and corresponding shear strength parameters were assigned to each rock layer. By setting the material constitutive model and interstrip force functions, setting the model boundary conditions based on topographic data and rock layer distribution data, and setting the potential sliding surface as the sliding surface along the basement interface, a landslide physical model is obtained.

[0011] Furthermore, the shear strength parameters of each sampling point under both states are input into the physical model to obtain the safety factor of each sampling point under both states, and compared with the preset threshold to obtain the parameter set of the natural stable state and the parameter set of the saturated unstable state. The specific steps are as follows: Assign the shear strength parameter of each acquisition point to the element in the physical model corresponding to the spatial location of that acquisition point. The shear strength parameters of each sampling point in its natural state are used as a set of input physical models to calculate the safety factor of the sampling point in its natural state; the shear strength parameters of each sampling point in its saturated state are used as a set of input physical models to calculate the safety factor of the sampling point in its saturated state. The safety factor of each sampling point under natural conditions is compared with the safety threshold. The shear strength parameters corresponding to sampling points with safety factors greater than the safety threshold are classified into the natural steady state parameter group. The safety factor of each sampling point under saturation is compared with the safety threshold. The shear strength parameters corresponding to sampling points with a safety factor less than the safety threshold are classified into the saturated unstable state parameter group.

[0012] Furthermore, a joint reliability evaluation index is obtained under the natural stable state and the saturated unstable state. The specific steps are as follows: The shear strength parameter data were divided into two groups: one group corresponds to the parameter data of the natural stable state, and the other group corresponds to the parameter data of the saturated unstable state. For these two sets of data, the probability distribution types of each parameter, including density, cohesion, and internal friction angle, are identified. The probability distribution types include normal distribution and log-normal distribution. For parameter data in both the naturally stable state and the saturated unstable state, the mean and standard deviation of specific gravity, cohesion, and internal friction angle are calculated respectively. For parameters identified as log-normally distributed, the logarithmic mean and logarithmic standard deviation are calculated after taking the logarithm of the original data. For the natural steady state: Based on the probability distribution type of each parameter under the natural steady state, for parameters that follow a normal distribution, their mean and standard deviation are directly used to generate normal random numbers. For parameters that follow a log-normal distribution, their logarithmic mean and standard deviation are used to generate log-normal random numbers. The Monte Carlo method is used for random sampling to generate N sets of shear strength parameter combinations. Each combination contains a unit weight value, a cohesion value, and an internal friction angle value. For saturated unstable state: Based on the probability distribution type of each parameter under saturated unstable state, for parameters that follow a normal distribution, their mean and standard deviation are used to generate normal random numbers. For parameters that follow a log-normal distribution, their logarithmic mean and standard deviation are used to generate log-normal random numbers. The Monte Carlo method is used for random sampling to generate N sets of shear strength parameter combinations. Each set of combinations includes a unit weight value, a cohesive force value, and an internal friction angle value. The generated N sets of shear strength parameters for the natural steady state are successively substituted into the physical model, and the safety factor corresponding to each set of parameters is calculated to obtain N safety factor values ​​for the natural steady state. The generated N sets of shear strength parameters for the saturated unstable state are successively substituted into the physical model, and the safety factor corresponding to each set of parameters is calculated by the limit equilibrium method to obtain N safety factor values ​​for the saturated unstable state. Based on the N safety factor values ​​under the natural steady state, statistical analysis is performed to calculate the joint reliability evaluation index under the natural steady state. The joint reliability evaluation index includes the average safety factor, the standard deviation of the safety factor, the instability probability, and the reliability index. Based on the N safety factor values ​​under saturated unstable state, statistical analysis is performed to calculate the joint reliability evaluation index under saturated unstable state. The joint reliability evaluation index includes the average safety factor, the standard deviation of the safety factor, the instability probability, and the reliability index.

[0013] Furthermore, the weighted reliability evaluation indexes for the naturally stable state and the saturated unstable state are obtained. The specific steps are as follows: For both the naturally stable state and the saturated unstable state, the average value of each parameter is used as the baseline value. Within a 100% range, the value of one parameter is changed successively while keeping the values ​​of other parameters unchanged. The combination of parameters after each change is substituted into the physical model to calculate the corresponding safety factor. The sensitivity change curves of each parameter are plotted based on the rate of change of the safety factor. The degree of influence of each parameter on landslide stability is determined by the absolute value of the slope of the curve. Based on the degree of influence of each parameter, the weighting coefficients of specific gravity, cohesion and internal friction angle are calculated using an improved three-scale analytic hierarchy process. For density, cohesion, and internal friction angle, based on the probability distribution type of the parameter under the corresponding state and the mean and standard deviation of the parameter under the corresponding state, the Monte Carlo method is used to randomly sample and generate multiple sets of random values ​​for the parameter. Each set of random values ​​is substituted into the physical model in turn to calculate the safety factor. At the same time, the other two parameters are fixed as the average value. The single reliability index of the parameter is obtained by statistical analysis based on the calculated safety factor value. The weighted reliability index is obtained by multiplying the weighting coefficients of each parameter by the individual reliability index of each parameter and then summing the results. Repeat the above steps for both the naturally stable state and the saturated unstable state to obtain the weighted reliability index for the naturally stable state and the weighted reliability index for the saturated unstable state.

[0014] Furthermore, the weighting coefficients for specific gravity, cohesion, and internal friction angle are calculated using an improved three-scale analytic hierarchy process. The specific steps are as follows: Based on the ranking of the influence of density, cohesion, and internal friction angle on landslide stability under natural conditions, a three-scale judgment matrix A is constructed, where the matrix contains the... Line 1 Column elements Values ​​are determined according to the following rules: When the The parameter is compared to the first When a parameter is important The value is 1; When the The parameter is compared to the first When both are equally important, The value is 0; When the The parameter is compared to the first When a parameter is important The value is -1; Based on the three-scale judgment matrix A, calculate the optimal transfer matrix B, where the matrix contains the first... Line 1 Column elements Equal to the first in matrix A Line 1 The average of the sums of corresponding elements in a column; Based on the optimal transfer matrix B, calculate the judgment matrix C, where the first... No. Column elements Equal to the natural constant e as the base, The power of the exponent; Based on the judgment matrix C, calculate the set average of all elements in each row to obtain the initial weights of each parameter. Then normalize each initial weight so that the sum of the weight coefficients of the three parameters equals 1, and obtain the final weight coefficients of specific gravity, cohesion and internal friction angle. Based on the degree of influence of unit weight, cohesion, and internal friction angle on landslide stability under saturated unstable conditions, a three-scale judgment matrix A is constructed. The matrix contains the first... Line 1 Column elements Values ​​are determined according to the following rules: When the The parameter is compared to the first When a parameter is important The value is 1; When the The parameter is compared to the first When both are equally important, The value is 0; When the The parameter is compared to the first When a parameter is important The value is -1; Based on the three-scale judgment matrix A, calculate the optimal transfer matrix B, where the matrix contains the first... Line 1 Column elements Equal to the first in matrix A Line 1 The average of the sums of corresponding elements in a column; Based on the optimal transfer matrix B, calculate the judgment matrix C, where the first... No. Column elements Equal to the natural constant e as the base, The power of the exponent; Based on the judgment matrix C, calculate the set average of all elements in each row to obtain the initial weights of each parameter. Then normalize each initial weight so that the sum of the weight coefficients of the three parameters equals 1, thus obtaining the final weight coefficients of density, cohesion and internal friction angle.

[0015] Furthermore, the joint reliability evaluation index and weighted reliability evaluation index for the two states are compared with preset accuracy thresholds and efficiency thresholds, respectively. The specific steps are as follows: Using the reliability index in the weighted reliability evaluation index as a benchmark, the accuracy threshold is set to k times the benchmark value, where k is a preset coefficient greater than 1. Using the calculation time of the joint reliability evaluation index as a benchmark, the efficiency threshold is set as 1 / m of the benchmark value, where m is a preset coefficient greater than 1. The reliability index in the joint reliability evaluation index is compared with the accuracy threshold. If the reliability index reaches or exceeds the accuracy threshold, the joint reliability evaluation index is determined to meet the accuracy threshold requirement. The actual calculation time of the weighted reliability evaluation index is compared with the efficiency threshold. If the calculation time is less than or equal to the efficiency threshold, the weighted reliability evaluation index is determined to meet the efficiency threshold requirement. If the joint reliability assessment index meets the accuracy threshold requirement, then the joint reliability assessment index is selected as the final reliability assessment index. If the joint reliability assessment index does not meet the accuracy threshold requirement, but the weighted reliability assessment index meets the efficiency threshold requirement, then the weighted reliability assessment index shall be selected as the final reliability assessment index. If the joint reliability assessment index does not meet the accuracy threshold requirement, and the weighted reliability assessment index does not meet the efficiency threshold requirement, then the weighted reliability assessment index shall be selected as the final reliability assessment index.

[0016] This invention also provides a dynamic assessment system for soil-rock composite landslides based on multi-source data fusion. The assessment system is used to execute the aforementioned dynamic assessment method for soil-rock composite landslides based on multi-source data fusion, and includes: The model building module sets up several collection points in the target landslide area and obtains the shear strength parameters of each collection point under natural and saturated conditions. It then constructs a physical model based on the landslide's topography and rock layer distribution characteristics. The parameter grouping module inputs the shear strength parameters of each sampling point under two states into the physical model to obtain the safety factor of each sampling point under the two states. Based on the safety factor, the stable state is determined. The shear strength parameters of the sampling points that are determined to be stable under the natural state are counted as the natural stable state parameter group, and the shear strength parameters of the sampling points that are determined to be unstable under the saturated state are counted as the saturated unstable state parameter group. The joint evaluation module determines the probability distribution type of each shear strength parameter in each of the natural stable state parameter group and the saturated unstable state parameter group. Monte Carlo random sampling is performed according to the probability distribution type. The generated new shear strength parameter group is input into the physical model to calculate the corresponding safety factor sample value. The joint reliability evaluation index under the two states is obtained based on the statistical results of all safety factor sample values. The weighted evaluation module determines the weight coefficient of each parameter based on the degree of influence of each shear strength parameter on landslide stability for the natural stable state parameter group and the saturated unstable state parameter group. It also calculates the individual reliability evaluation index for each shear strength parameter and sums the weights with the corresponding individual reliability evaluation indexes to obtain the weighted reliability evaluation indexes for the two states. The decision optimization module compares the joint reliability evaluation index and the weighted reliability evaluation index of the two states with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, the joint reliability evaluation index is selected as the final reliability evaluation index.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention introduces parameter state dependence and divides the shear strength parameter into two states, natural stability and saturated instability, for separate evaluation, thus achieving a breakthrough in landslide reliability evaluation from static analysis to dynamic assessment. This invention integrates three uncertainty modeling methods: distribution type identification, spatial variability quantification, and parameter correlation analysis, making the parameter statistical characteristics more consistent with reality and significantly improving the accuracy of the evaluation results. This invention designs two modes: joint calculation and weighted calculation, and establishes a decision-making mechanism based on accuracy and efficiency thresholds. It can automatically select the best evaluation results according to engineering needs, taking into account both evaluation accuracy and computational efficiency. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a flowchart of the method steps of the present invention; Figure 3 This is a block diagram of the overall system structure of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: Please see Figure 1-3 The present invention provides a technical solution: The dynamic assessment method for soil-rock composite landslides based on multi-source data fusion includes the following steps: Step 1: Set up several sampling points in the target landslide area, and obtain the shear strength parameters of each sampling point under natural and saturated conditions. Construct a physical model based on the topography and rock layer distribution characteristics of the landslide.

[0022] In this embodiment, in order to accurately characterize the overall stability of the target landslide area, a typical exploration profile must first be determined in the target landslide area. The selection of this profile must follow the following principles: its orientation must be consistent with the main sliding direction of the landslide, and the profile line should penetrate as much as possible through the key parts of the landslide body, specifically including the rear edge, middle and front edge of the landslide body. The purpose of this is to ensure that the soil and rock samples collected on this profile and the analysis conducted can represent the deformation and failure characteristics and mechanical behavior of the entire landslide body to the greatest extent, and to avoid systematic deviations in subsequent evaluation results due to improper profile selection.

[0023] After identifying a typical survey profile, sampling points were set up along this profile, employing a combination of evenly spaced points and increased density at key locations. In this embodiment, sampling points were evenly distributed along the profile line from the rear edge to the front edge of the landslide, with a basic spacing of 10 meters. Simultaneously, sampling points were increased at key locations such as areas with drastic changes in terrain slope, areas with existing cracks, areas with significant variations in the depth of the subsurface interface, and groundwater outcrops, with the spacing adjusted to 5 meters. This layout ensures both uniform coverage of the entire profile and the ability to effectively capture vulnerable areas that may control landslide stability.

[0024] For natural conditions, during the dry season or periods without rainfall, undisturbed soil samples are taken at designated collection points to obtain soil and rock samples of the landslide body under normal natural conditions.

[0025] To assess the instability risk under extreme rainfall conditions, after sampling in the natural state, and following one effective rainfall event during the rainy season, soil samples were taken again from a nearby location on the third day after the rainfall ceased. The reason for choosing the third day after rainfall cessation was that preliminary investigations and tests of the permeability characteristics of the soil and rock mass in the target landslide area confirmed that the rainwater infiltration rate and the time required for the wetting front to migrate to the sliding surface depth of the landslide's sliding layer soil and rock mass were approximately three days. At this point, the rainwater had fully infiltrated to the sliding surface depth, most accurately reflecting the most adverse impact of rainfall on the shear strength of the soil and rock mass. It should be noted that for soil and rock masses with different permeability characteristics, the sampling window period should be adjusted accordingly based on the wetting front migration time determined by field infiltration tests or empirical formulas, using the depth of the infiltration front reaching the sliding surface as the sampling time benchmark to ensure the representativeness and accuracy of the saturated state sampling.

[0026] To ensure the representativeness of the collected samples, the sampling depth at each sampling point should at least penetrate the sliding layer and reach or approach the basement interface, so as to ensure that the obtained parameters can reflect the mechanical properties of the soil and rock mass near the potential sliding surface.

[0027] To meet the basic requirements for data volume in subsequent large-sample statistical analysis and to provide a reliable statistical basis for modeling parameter uncertainty, the number of valid soil and rock samples collected independently in both natural and saturated states shall not be less than 38.

[0028] The collected soil and rock samples were sealed and preserved, and promptly transported to the laboratory for indoor geotechnical testing. Three core shear strength parameters were obtained for each sample at each collection point: unit weight, cohesion, and angle of internal friction. For each condition, at least 38 measured data points for each parameter were obtained, forming the original data sequence.

[0029] Based on the shear strength parameter data, further topographic and geomorphological data and rock strata distribution data of the target area are collected. The topographic and geomorphological data are derived from the topographic elevation information corresponding to typical exploration profiles. The rock strata distribution data are obtained in the following ways: combining regional geological data and on-site engineering geological exploration, the thickness, occurrence, distribution range and interlayer contact relationship of each rock stratum are determined by means of drilling, trenching and geophysical exploration, and engineering geological profile maps are drawn.

[0030] Based on the above data, a physical model of the landslide was established. The specific modeling process is as follows: Establish a geometric model: Based on the topographic data of typical survey profiles, establish a two-dimensional limit equilibrium analysis geometric model of the landslide. Import the topographic elevation data corresponding to the survey profile line into the modeling software to generate a two-dimensional profile model that is completely corresponding to the longitudinal main profile in the same direction as the main sliding direction. Stratigraphic Unit Division: Based on rock strata distribution data, the landslide rock strata sequence is divided into two main stratigraphic units from top to bottom, according to the engineering geological profile: the overlying sliding layer and the underlying non-sliding bedrock layer. The overlying sliding layer is the soil and rock mass that may undergo sliding failure, mainly composed of Quaternary loose deposits; the underlying non-sliding bedrock layer is stable bedrock that will not undergo sliding failure under normal working conditions. Assigning material properties: After layering, shear strength parameters corresponding to the respective states are assigned to each rock layer. The unit weight, cohesion, and internal friction angle data are assigned to the overlying sliding layer according to its natural and saturated states, respectively. The underlying non-sliding bedrock layer is assigned corresponding strength parameters based on its lithological characteristics. Simultaneously, a material constitutive model and interstrip force functions are defined. The Mohr-Coulomb model is used for the material constitutive model to describe the stress-strain relationship and failure criterion of the soil and rock mass; the half-sine function is used for the interstrip force functions to describe the distribution of forces between soil strips in subsequent stability calculations. This completes the construction of the mechanical properties of the physical model. Setting boundary conditions: Based on terrain and rock layer distribution data, the model boundary conditions are set, specifically including: the bottom of the model is set as a completely fixed constraint to simulate the stable state of deep bedrock; the two sides of the model are set as horizontal constraints, allowing vertical displacement to simulate the stress state of the strata in a semi-infinite space. This completes the construction of the boundary conditions for the physical model. Define the potential sliding surface: The potential sliding surface is defined as the sliding surface along the interface between the substrate and the overlying layer. In the subsequent stability calculation, the sliding trend of the overlying sliding layer along this interface is mainly considered.

[0031] Step 2: Input the shear strength parameters of each sampling point under the two states into the physical model to obtain the safety factor of each sampling point under the two states. Determine the stable state based on the safety factor. Collect the shear strength parameters of the sampling points determined to be stable under the natural state as the natural stable state parameter group, and collect the shear strength parameters of the sampling points determined to be unstable under the saturated state as the saturated unstable state parameter group.

[0032] In the two-dimensional limit equilibrium analysis geometric model, the overlying sliding layer is divided into several continuous computational units along the profile line. Each unit has a defined location range. Based on the horizontal distance of each sampling point, the range of which computational unit the sampling point falls within is determined. For each sampling point, the unit containing the coordinates of that location is found in the divided computational grid, and the measured shear strength parameter of that sampling point is assigned to that unit. When multiple sampling points fall within the same unit range, the average value of the parameters of these sampling points is taken as the parameter value of that unit. When there are no sampling points within a unit range, the parameter values ​​of adjacent units are used to determine the parameter value through linear interpolation. Through the above operations, the parameter settings of all computational units are completed, enabling the physical model to reflect the spatial distribution characteristics of the shear strength parameter along the profile.

[0033] After establishing the spatial correspondence between parameters and the model, for the natural state, the shear strength parameters measured at each sampling point under natural conditions, including unit weight, cohesion, and internal friction angle, are assigned to the corresponding computational units in the physical model according to the aforementioned spatial correspondence rules. Then, the limit equilibrium method is used to calculate stability, obtaining the safety factor of the local area represented by that sampling point under natural conditions. This process is repeated, iterating through all sampling points, to obtain a sequence of safety factors under natural conditions.

[0034] For the saturated state, the same method is used, treating the shear strength parameters measured at each sampling point under saturated conditions as another set of independent input parameters. These parameters are then assigned to the corresponding computational units in the physical model according to the same spatial correspondence rules, thus calculating the safety factor for that sampling point under saturated conditions. After traversing all sampling points, a sequence of safety factors under saturated conditions is obtained.

[0035] According to the principle of limit equilibrium, the safety factor is defined as the ratio of the anti-slip force to the sliding force on the sliding surface. In this embodiment, the safety threshold is 1. When the safety factor equals 1, the anti-slip force and the sliding force are equal, and the landslide is in a state of limit equilibrium; when the safety factor is greater than 1, the anti-slip force is greater than the sliding force, and the landslide is in a stable state; when the safety factor is less than 1, the anti-slip force is less than the sliding force, and the landslide is in an unstable state.

[0036] Construction of the natural stable state parameter group: The safety factor of each sampling point under the natural state is compared with the threshold 1. If the safety factor under the natural state is greater than 1, it indicates that the landslide as a whole is in a stable state under the local soil and rock parameters represented by that sampling point. This set of shear strength parameters can reflect the soil and rock mechanical properties under stable working conditions. The natural state shear strength parameters corresponding to that sampling point, including unit weight, cohesion, and internal friction angle, are included in the natural stable state parameter group.

[0037] Construction of the saturated unstable state parameter group: The safety factor of each sampling point under saturated state is compared with the threshold 1. If the safety factor under saturated state is less than or equal to 1, it indicates that the landslide as a whole is in an unstable state under the local soil and rock parameters represented by that sampling point. This group of shear strength parameters can reflect the soil and rock mechanical properties under the unstable condition. The saturated state shear strength parameters corresponding to that sampling point, including unit weight, cohesion, and internal friction angle, are included in the saturated unstable state parameter group.

[0038] Through the above grouping, the originally collected samples were divided into two parameter groups with clear physical meaning: the natural stable state parameter group and the saturated unstable state parameter group. It should be noted that the natural state parameters and saturated state parameters at the same collection point may belong to different parameter groups, reflecting the characteristic that shear strength parameters undergo state transitions with changes in the external environment.

[0039] Step 3: For the natural stable state parameter group and the saturated unstable state parameter group, determine the probability distribution type of each shear strength parameter in each group, perform Monte Carlo random sampling according to the probability distribution type, input the generated new shear strength parameter group into the physical model, calculate the corresponding safety factor sample value, and obtain the joint reliability evaluation index under the two states based on the statistical results of all safety factor sample values.

[0040] The natural stable state parameter set and the saturated unstable state parameter set are defined as two sets of input data. One set is the natural stable state parameter set, which contains the measured data of density, cohesion and internal friction angle of all sampling points under this state; the other set is the saturated unstable state parameter set, which contains the measured data of density, cohesion and internal friction angle of all sampling points under this state.

[0041] For each set of data, the probability distribution type of the three parameters—density, cohesion, and internal friction angle—was identified. The probability distribution types include two basic types: normal distribution and log-normal distribution. The identification of the distribution type employed a dual-verification method combining the Shapiro-Wilk test and the QQ plot: First, the measured data for each parameter underwent a Shapiro-Wilk normality test, with a significance level of 0.05 as the criterion. If the p-value was greater than 0.05, it was preliminarily determined that the distribution followed a normal distribution; otherwise, it was preliminarily determined that the distribution did not follow a normal distribution. Then, the QQ plot was used for final confirmation. If the sample scatter points in the QQ plot were basically distributed along the theoretical straight line, it was confirmed that the distribution followed a normal distribution; if there was a significant deviation, it was determined to be a log-normal distribution.

[0042] After determining the probability distribution type of each parameter, the statistical characteristic value of each parameter is calculated for the natural stable state parameter group and the saturated unstable state parameter group respectively. For parameters identified as following a normal distribution, the mean and standard deviation are directly calculated based on the measured data of the parameter group. For parameters identified as following a log-normal distribution, the natural logarithm of the original measured data is first taken to obtain a set of log-transformed data sequences, and then the logarithmic mean and logarithmic standard deviation are calculated based on this set of logarithmic data.

[0043] Based on the probability distribution types and statistical characteristics of the parameters determined above, Monte Carlo random sampling was performed for both the naturally stable state and the saturated unstable state.

[0044] For the naturally stable state, random numbers are generated based on the distribution type and statistical characteristics of each parameter under this state. For parameters following a normal distribution, their mean and standard deviation are directly used to generate a random number sequence conforming to a normal distribution; for parameters following a log-normal distribution, their logarithmic mean and logarithmic standard deviation are used to generate a random number sequence conforming to a log-normal distribution. Random numbers for specific gravity, cohesion, and internal friction angle are independently extracted and combined to generate N sets of shear strength parameter combinations. Each combination contains one specific gravity value, one cohesion value, and one internal friction angle value. In this embodiment, N is taken as 10,000 times.

[0045] For the saturated unstable state, the same method is used to independently generate N sets of shear strength parameter combinations based on the distribution type and statistical characteristics of each parameter under this state.

[0046] The two sets of shear strength parameter combinations generated above are substituted into the physical model for stability calculation. For the natural steady state, the ten thousand sets of shear strength parameter combinations generated are sequentially input into the physical model. For each set of parameters, the spatial correspondence rule described in step 2 is used to assign it to the corresponding calculation unit in the model, and the safety factor corresponding to that set of parameters is calculated. This process is repeated until all ten thousand sets of parameter combinations are traversed, thereby obtaining ten thousand sample values ​​of safety factors under the natural steady state.

[0047] For the saturated unstable state, the same method is used to sequentially input the generated 10,000 sets of shear strength parameter combinations into the physical model, calculate the safety factor corresponding to each set of parameters, and obtain 10,000 safety factor sample values ​​under the saturated unstable state.

[0048] Based on the obtained safety factor sample values, statistical analysis was performed on both the naturally stable state and the saturated unstable state to calculate the joint reliability assessment index. The joint reliability assessment index includes the average safety factor, the standard deviation of the safety factor, the instability probability, and the reliability index. The average safety factor is obtained by summing all safety factor sample values ​​and dividing by the total number of samples. The standard deviation of the safety factor is calculated using the sample standard deviation, which is the sum of the squared differences between each safety factor sample value and the average safety factor, divided by the total number of samples minus one, and finally the square root. The instability probability is calculated by counting the number of samples with a safety factor less than 1 under that state, dividing that number by the total number of samples, and the probability ranges from 0 to 1; a higher value indicates a higher probability of landslide instability. The reliability index is calculated by subtracting 1 from the average safety factor and dividing by the standard deviation of the safety factor. A positive reliability index indicates a stable landslide, with a higher value indicating better stability; a negative reliability index indicates an unstable landslide.

[0049] In this embodiment, for the K156+200 to +800 landslide, the natural stable state parameter set and the saturated unstable state parameter set obtained through the above steps are used as inputs for this step. Based on distribution type identification, in the natural stable state, the density and internal friction angle follow a normal distribution, and the cohesion follows a log-normal distribution; in the saturated unstable state, the density, cohesion, and internal friction angle all follow a log-normal distribution. Based on the above distribution types and their statistical characteristics, ten thousand Monte Carlo random samples are performed for both states. The generated parameter combinations are then substituted into the physical model for calculation. Statistical analysis yields the following joint reliability evaluation indexes for the natural stable state: average safety factor 1.133, standard deviation of safety factor 0.017, instability probability 0%, and reliability index 7.824; and for the saturated unstable state: average safety factor 0.855, standard deviation of safety factor 0.017, instability probability 100%, and reliability index -8.529.

[0050] To verify the effectiveness of the parameter true distribution identification method used in this invention, the calculation results based on the true distribution (condition 1) are compared with two common simple assumption conditions. Condition 2 assumes that all parameters follow a normal distribution, and condition 3 assumes that all parameters follow a log-normal distribution. Both comparison conditions ignore the spatial variability and correlation of the parameters. The comparison of the calculation results for the three conditions is shown in Table 1.

[0051] Table 1: Comparison of Joint Reliability Assessment Results under Different Probability Distribution Assumptions Table 1 compares the joint reliability assessment results under different probability distribution assumptions. The comparison in Table 1 shows that, regarding the reliability index, under the natural steady state, the reliability index of Case 1 is 7.824, significantly higher than Case 2's 2.281 and Case 3's 3.047; under the saturated unstable state, the reliability index of Case 1 is -8.529, and its absolute value is also much higher than Case 2's -2.917 and Case 3's -4.086. This indicates that using simple normal or log-normal assumptions will severely underestimate the stability level of the landslide (natural state) or overestimate its safety (saturated state), with a deviation exceeding 60%.

[0052] Regarding the standard deviation of the safety factor, under the naturally stable state, the standard deviation of condition 1 is 0.017, which is much smaller than that of condition 2 (0.057) and condition 3 (0.043). Under the saturated unstable state, the standard deviation of condition 1 is also 0.017, which is smaller than that of condition 2 (0.048) and condition 3 (0.035). This indicates that by identifying the true distribution type of the parameters, the present invention significantly reduces the dispersion of the evaluation results and achieves higher stability.

[0053] Regarding the probability of instability, Case 1 has a 0% probability of instability under naturally stable conditions and a 100% probability under saturated unstable conditions, with a clear judgment result. However, Cases 2 and 3 still have a 0.93% probability of misjudging instability under naturally stable conditions, and also fail to achieve a 100% complete instability judgment under saturated unstable conditions. This indicates that incorrect distribution assumptions can lead to misjudgments of instability risk.

[0054] Step 4: For the natural stable state parameter group and the saturated unstable state parameter group, determine the weight coefficient of each parameter according to the degree of influence of each shear strength parameter on landslide stability in each group, and calculate the individual reliability assessment index of each shear strength parameter. The weight and the corresponding individual reliability assessment index are weighted and summed to obtain the weighted reliability assessment index for the two states.

[0055] After completing the above-mentioned joint reliability assessment, this step provides another assessment mode: single-parameter weighted calculation. For both the naturally stable and saturated unstable states, parameter sensitivity analysis is performed to determine the order of influence of the three parameters—unit weight, cohesion, and internal friction angle—on landslide stability. The specific operation method for sensitivity analysis is as follows: using the average value of each parameter under this state as the baseline value, within a range of plus or minus 100% of the baseline value, the value of one parameter is successively changed while keeping the values ​​of the other two parameters fixed at the baseline value. The changed parameter combination is then substituted into the physical model, and the corresponding safety factor is calculated using the limit equilibrium method.

[0056] Based on the safety factor values ​​calculated above, sensitivity change curves for each parameter are plotted. The horizontal axis of the curve represents the magnitude of parameter change, and the vertical axis represents the corresponding safety factor value. By calculating the slope at each point on the curve and taking the absolute value of the slope, the degree of influence of the parameter change on the safety factor can be quantitatively measured. The larger the absolute value of the curve slope, the more significant the impact of the parameter on landslide stability; the smaller the absolute value of the curve slope, the weaker the impact of the parameter.

[0057] Based on the absolute value of the slope of the sensitivity curves of each parameter, the degree of influence of specific gravity, cohesion, and internal friction angle are ranked to obtain the parameter importance sequence.

[0058] Based on the parameter influence ranking obtained from the above sensitivity analysis, an improved three-scale analytic hierarchy process (AHP) is used to calculate the weight coefficients of density, cohesion, and internal friction angle. This method avoids the problems of difficult construction and complex consistency checks of the nine-scale judgment matrix in traditional AHP by constructing and transforming a three-scale judgment matrix.

[0059] The steps for calculating the weighting coefficients are as follows: The first step is to construct a three-scale judgment matrix. Based on the ranking of parameter influence, a three-row, three-column judgment matrix is ​​constructed, where the first parameter is the first parameter. Line number The rule for the values ​​of the elements in a column is: when the first element... The parameter is compared to the first When the first parameter is important, the element takes the value 1; when the second parameter is important, the element takes the value 1. The parameter and the first When all parameters are equally important, this element takes the value 0; when the first parameter is of equal importance... The parameter is compared to the first When a parameter is important, the element takes the value of -1.

[0060] The second step is to calculate the optimal transfer matrix. Based on the three-scale judgment matrix mentioned above, the optimal transfer matrix is ​​calculated. The optimal transfer matrix contains the first... Line number The elements of the column are equal to the elements of the three-scale judgment matrix. All elements in the row and the first The average of the sums of all corresponding elements in the column.

[0061] The third step is to calculate the judgment matrix. Based on the optimal transit matrix, calculate the judgment matrix, where the first... Line number The elements of a column are equal to powers of the natural constant e, with the corresponding element in the optimal transfer matrix as the exponent.

[0062] The fourth step is to calculate and normalize the weight coefficients. Based on the judgment matrix, calculate the geometric mean of all elements in each row to obtain the initial weights of each parameter. Then, add the three initial weights together to get the total. Divide each initial weight by the total to make the sum of the weight coefficients of the three parameters equal to 1, thus obtaining the final weight coefficients of specific gravity, cohesion, and internal friction angle.

[0063] For both the naturally stable state and the saturated unstable state, the weighting coefficients are calculated independently according to the steps described above.

[0064] After obtaining the weighting coefficients for each parameter, it is necessary to calculate the individual reliability index for each parameter. The individual reliability index reflects the degree of influence of the uncertainty of a single parameter on landslide stability.

[0065] The calculation method for the single reliability index is as follows: For the three parameters—specific gravity, cohesion, and internal friction angle—random sampling is performed using the Monte Carlo method, based on the probability distribution type of the parameter under the corresponding state and the mean and standard deviation of the parameter under the corresponding state. During sampling, only one random value for the parameter is generated each time, while the other two parameters are fixed at their respective average values. The generated random value for the parameter is combined with the other two fixed parameters and substituted into the physical model to calculate the safety factor. The above process is repeated for a large number of samples to obtain a series of safety factor sample values ​​corresponding to the parameter.

[0066] Based on the 10,000 safety factor sample values ​​obtained above, statistical analysis is performed to calculate the individual reliability index of this parameter, specifically including the calculation of the following four indicators: The average safety factor is calculated as follows: add up all 10,000 safety factor sample values, divide by the total number of samples (10,000), and obtain the average safety factor. The standard deviation of the safety factor is calculated as follows: square the difference between each safety factor sample value and the average safety factor, add all the squared values ​​together, divide by the total number of samples (ten thousand minus one), and then take the square root of the result to obtain the standard deviation of the safety factor. The instability probability is calculated as follows: count the number of samples with a safety factor less than 1 in 10,000 safety factor samples, divide the number by the total number of samples (10,000) to obtain the instability probability; The reliability index is calculated by subtracting 1 from the average safety factor and then dividing by the standard deviation of the safety factor.

[0067] The above four indicators together constitute the single reliability evaluation index of this parameter, with the reliability index serving as the main input for subsequent weighted calculations.

[0068] After obtaining the weighting coefficients and individual reliability indices of each parameter, the weighted reliability index is calculated by linear weighted summation. The calculation method of the weighted reliability index is as follows: multiply the weighting coefficient of severity by the individual reliability index of severity, multiply the weighting coefficient of cohesion by the individual reliability index of cohesion, and multiply the weighting coefficient of internal friction angle by the individual reliability index of internal friction angle. Then, add the three products together, and the result is the weighted reliability index.

[0069] For the naturally stable state and the saturated unstable state, the weighted reliability index under the naturally stable state and the weighted reliability index under the saturated unstable state are calculated independently according to the above method.

[0070] The single-parameter weighted calculation mode provided in this step complements the multi-parameter joint calculation mode in the above steps. When engineering survey data is limited and cannot meet the requirements of large-sample statistical analysis, or when the project has high requirements for calculation efficiency, the weighted calculation method in this step can significantly reduce the calculation complexity and time cost while ensuring a certain level of accuracy.

[0071] In this embodiment, single-parameter weighted reliability assessments were performed on the K156+200 to +800 landslide under both the naturally stable and saturated unstable states. Sensitivity analysis results show that under the naturally stable state, the parameter sensitivity ranking is: cohesion has the greatest impact, followed by the internal friction angle, and then the density has the least impact; under the saturated unstable state, the parameter sensitivity ranking is: cohesion has the greatest impact, followed by the internal friction angle, and then the density has the least impact. The sensitivity ranking is consistent in both states, but the weighting coefficients are calculated independently.

[0072] Under natural stable conditions, based on the ranking result of cohesion > internal friction angle > specific weight, a three-scale judgment matrix A is constructed. The rule for the three scale values ​​is: when the... The parameter is compared to the first Each parameter is assigned a value of 1 when it is important, 0 when they are equally important, and -1 when they are less important than the other. The resulting judgment matrix A is: first row (cohesion) [0, 1, 1], second row (internal friction angle) [-1, 0, 1], third row (weight) [-1, -1, 0]. Based on this, following the calculation process from steps one to four, the optimal transfer matrix B and judgment matrix C are calculated sequentially. Finally, the weight coefficients in the natural stable state are normalized to obtain: cohesion 0.563, internal friction angle 0.289, weight 0.149, and the sum of the three terms is 1.001. Under saturated unstable conditions, although the sensitivity ranking is the same as in the natural state (cohesion > internal friction angle > specific gravity), the deterioration effect of specific gravity is significantly increased in the saturated state, and its influence on landslide stability is relatively enhanced. Therefore, the constructed three-scale judgment matrix A' is as follows: first row (cohesion) [0, 1, 1], second row (internal friction angle) [-1, 0, 0], third row (specific gravity) [-1, 0, 0]. Compared with the judgment matrix A in the natural state, the elements in the second row and third column (internal friction angle relative to specific gravity) of A' are adjusted from 1 to 0, and the elements in the third row and second column (specific gravity relative to internal friction angle) are adjusted from -1 to 0, reflecting the relatively enhanced influence of specific gravity in the saturated state. Based on this, following the calculation process from step one to step four mentioned above, the optimal transfer matrix B and the judgment matrix C are calculated sequentially. Finally, the weight coefficients under the natural stable state are normalized to obtain: cohesion 0.520, internal friction angle 0.260, specific weight 0.220, and the sum of the three terms is 1.000.

[0073] In the calculation of the individual reliability index, ten thousand Monte Carlo samplings were performed for each parameter to obtain the individual reliability index for each parameter. Under the natural stable state, the individual reliability index for heavy-duty reliability is 5.221, the cohesion is 9.042, and the internal friction angle is 10.842; under the saturated unstable state, the index for heavy-duty reliability is -20.811, the cohesion is -15.386, and the internal friction angle is -11.468.

[0074] The weighted sum of the weighting coefficients and the individual reliability indices yields a weighted reliability index of 9.002 under the naturally stable state and a weighted reliability index of -15.077 under the saturated unstable state.

[0075] Table 2: Summary Table of Single-Parameter Weighted Reliability Assessment Results Table 2 summarizes the results of the single-parameter weighted reliability assessment. A comparison in Table 2 shows that, regarding the weighting coefficients, the weight of cohesion is 0.563 in both states, significantly higher than the 0.289 for the internal friction angle and 0.149 for the unit weight. This is entirely consistent with the sensitivity analysis results, indicating that cohesion is the dominant parameter controlling the stability of this landslide and should be a key focus in prevention and control design.

[0076] Regarding individual reliability indices, all three parameters showed positive values ​​under the natural stable state, with the internal friction angle having the highest value (10.842), followed by cohesion (9.042), and the heaviest (5.221). Under the saturated unstable state, all three parameters showed negative values, with the heaviest value having the largest absolute value (-20.811), followed by cohesion (-15.386), and the internal friction angle having the smallest (-11.468). This pattern indicates that the risk contribution of different parameters varies significantly under different states: the internal friction angle contributes the most under the natural state, while the heaviest deterioration effect is most significant under the saturated state.

[0077] Regarding the average safety factor, under the natural steady state, the average safety factor of each parameter acting alone is greater than 1, ranging from 1.073 to 1.122; under the saturated unstable state, the average safety factor is less than 1, ranging from 0.810 to 0.846. This further confirms the decisive influence of parameter state transition on landslide stability.

[0078] Regarding the standard deviation of the safety factor, the standard deviations of each parameter under natural steady-state conditions range from 0.005 to 0.011, while those under saturated unstable conditions range from 0.003 to 0.011. Among them, the standard deviation of severe conditions is the smallest (0.005 for natural conditions and 0.003 for saturated conditions), indicating that the variability of the severe parameters is relatively low, and their uncertainty has a smaller impact on the assessment results.

[0079] Regarding the probability of instability, the probability of instability for each parameter is 0% under the natural stable state and 100% under the saturated unstable state. The judgment results are clear and consistent, indicating that each parameter can clearly reflect the stability properties of the landslide under different states when acting alone.

[0080] Table 3: Weighted Reliability Comprehensive Evaluation Results Table 3 shows the weighted reliability comprehensive assessment results. As can be seen from Table 3, the weighted reliability index under the natural stable state is 9.002, the weighted average safety factor is 1.109, and the instability probability is 0%, indicating that the landslide is stable under natural conditions. Under the saturated unstable state, the weighted reliability index is -15.077, the weighted average safety factor is 0.835, and the instability probability is 100%, indicating that the landslide has an extremely high risk of instability under saturated conditions. The weighted assessment results are consistent with the pattern of individual parameter assessments, and through the integration of weighting coefficients, comprehensively reflect the contribution of each parameter to landslide stability.

[0081] Step 5: Compare the joint reliability evaluation index and weighted reliability evaluation index of the two states with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, select the joint reliability evaluation index as the final reliability evaluation index.

[0082] The accuracy threshold is set based on the reliability index in the weighted reliability assessment index. Specifically, the weighted reliability index is multiplied by a preset coefficient k greater than 1, and the result is the accuracy threshold. The value of this preset coefficient k is determined according to the importance level of the project. For landslide projects with high importance, k can be taken as a smaller value to require higher accuracy; for general projects, k can be taken as a larger value to relax the accuracy requirements. The accuracy threshold is used to measure whether the accuracy of the joint reliability assessment index has reached an acceptable level.

[0083] The efficiency threshold is set based on the calculation time of the joint reliability assessment index. Specifically, the total calculation time required for the joint reliability assessment in step 3, from the start of Monte Carlo sampling to the output result, is recorded. This time is then divided by a preset coefficient m greater than 1, and the result is the efficiency threshold. The value of this preset coefficient m is determined according to the urgency of the project. For emergency scenarios requiring rapid decision-making, m can be taken as a smaller value to demand higher efficiency; for routine assessments, m can be taken as a larger value to relax efficiency requirements. The efficiency threshold is used to measure whether the calculation speed of the weighted reliability assessment index meets the timeliness requirements.

[0084] After setting the accuracy threshold and efficiency threshold, the two evaluation metrics are compared. The reliability index in the joint reliability evaluation metric is compared with the accuracy threshold. If the joint reliability index reaches or exceeds the accuracy threshold (i.e., the absolute value of the joint reliability index is greater than or equal to the absolute value of the accuracy threshold), the joint reliability evaluation metric is deemed to meet the accuracy threshold requirement; otherwise, it is deemed not to meet the accuracy threshold requirement.

[0085] The actual computation time consumed by the weighted reliability assessment is compared with the efficiency threshold. If the computation time of the weighted reliability assessment is less than or equal to the efficiency threshold, the weighted reliability assessment index is determined to meet the efficiency threshold requirement; otherwise, it is determined not to meet the efficiency threshold requirement.

[0086] Based on the above comparison results, the final reliability evaluation index will be selected according to the following rules: In the first scenario, if the joint reliability assessment index meets the accuracy threshold requirement, then regardless of whether the weighted reliability assessment index meets the efficiency threshold requirement, the joint reliability assessment index should be selected as the final reliability assessment index. This is because, under the premise of meeting the accuracy requirement, the comprehensiveness and robustness of the joint assessment results are superior, and it should be given priority.

[0087] In the second scenario, if the joint reliability assessment index does not meet the accuracy threshold requirement, but the weighted reliability assessment index meets the efficiency threshold requirement, then the weighted reliability assessment index is selected as the final reliability assessment index. This is because when the joint assessment accuracy is insufficient, it indicates that the current data conditions or model settings may not be suitable for complex joint calculations. In this case, using the weighted assessment result as an alternative can meet the efficiency requirements while ensuring a certain level of accuracy.

[0088] In the third scenario, if the joint reliability assessment index does not meet the accuracy threshold requirement, and the weighted reliability assessment index does not meet the efficiency threshold requirement, then the weighted reliability assessment index is still selected as the final reliability assessment index. This is because, in this case, although the efficiency of the weighted assessment does not meet expectations, it is still a relatively feasible assessment result and can serve as a reference for engineering decisions.

[0089] In this embodiment, decisions are made for the natural stable state and the saturated unstable state for the K156+200 to +800 landslide.

[0090] Based on the importance of the project, the preset coefficient k is set to 1.2. Under the naturally stable state, the weighted reliability index is 9.002, which, multiplied by 1.2, yields an accuracy threshold of 10.802. Under the saturated unstable state, the weighted reliability index is -15.077, which, multiplied by 1.2, yields an accuracy threshold of -18.092. Regarding the efficiency threshold, the calculation time for the joint reliability assessment is recorded as 120 minutes, and the preset coefficient m is set to 2. Dividing 120 minutes by 2 yields an efficiency threshold of 60 minutes.

[0091] Then, comparisons were made. Under the naturally stable state, the joint reliability index was 7.211. Compared with the accuracy threshold of 10.802, 7.824 is less than 10.802, thus failing to meet the accuracy threshold requirement. Under the saturated unstable state, the joint reliability index was -8.529. Compared with the accuracy threshold of -18.092, the absolute value of -8.529 is less than the absolute value of -18.092, again failing to meet the accuracy threshold requirement. In terms of efficiency, the actual calculation time for the weighted reliability assessment was 45 minutes. Compared with the efficiency threshold of 60 minutes, 45 minutes is less than 60 minutes, thus meeting the efficiency threshold requirement.

[0092] According to the decision-making rules, the joint reliability assessment index does not meet the accuracy threshold requirement, but the weighted reliability assessment index meets the efficiency threshold requirement. Therefore, the weighted reliability assessment index is selected as the final reliability assessment index for this landslide. The final assessment result under the natural steady state is a weighted reliability index of 9.002, and the final assessment result under the saturated unstable state is a weighted reliability index of -15.077.

[0093] Table 4: Comparison of Joint Evaluation and Weighted Evaluation Results Table 4 compares the results of the joint evaluation and the weighted evaluation. As shown in Table 4, regarding the reliability index, the joint evaluation result is 7.824 under the naturally stable state, while the weighted evaluation result is 9.002; under the saturated unstable state, the joint evaluation result is -8.529, while the weighted evaluation result is -15.077. The trends of the two evaluation methods are completely consistent, both indicating stability under the naturally stable state and instability under the saturated state, but there are numerical differences. This difference stems from the different calculation logics of the two methods: the joint evaluation comprehensively considers the correlation and spatial variability between parameters, while the weighted evaluation focuses more on the independent contribution of each parameter.

[0094] Regarding the average safety factor, the joint assessment under the naturally stable state is 1.133, slightly higher than the weighted assessment of 1.109; the joint assessment under the saturated unstable state is 0.855, slightly higher than the weighted assessment of 0.835. Both differences are within acceptable limits.

[0095] Regarding the probability of instability, the results of the two models are completely consistent: 0% under the natural stable state and 100% under the saturated unstable state, indicating that the qualitative judgment of the stability state of this landslide is consistent regardless of which calculation model is used.

[0096] Regarding the standard deviation of the safety factor, the joint assessment can output this indicator (0.017 for both natural and saturated values), while the weighted assessment, being a weighted summation, cannot directly calculate the comprehensive standard deviation and is therefore indicated by "—" in the table. This demonstrates the advantage of the joint assessment in terms of information richness.

[0097] Please see Figure 3 The present invention further provides a dynamic assessment system for soil-rock composite landslides based on multi-source data fusion. This assessment system is used to execute the aforementioned dynamic assessment method for soil-rock composite landslides based on multi-source data fusion, and includes: The model building module sets up several collection points in the target landslide area and obtains the shear strength parameters of each collection point under natural and saturated conditions. It then constructs a physical model based on the landslide's topography and rock layer distribution characteristics. The parameter grouping module inputs the shear strength parameters of each sampling point under two states into the physical model to obtain the safety factor of each sampling point under the two states. Based on the safety factor, the stable state is determined. The shear strength parameters of the sampling points that are determined to be stable under the natural state are counted as the natural stable state parameter group, and the shear strength parameters of the sampling points that are determined to be unstable under the saturated state are counted as the saturated unstable state parameter group. The joint evaluation module determines the probability distribution type of each shear strength parameter in each of the natural stable state parameter group and the saturated unstable state parameter group. Monte Carlo random sampling is performed according to the probability distribution type. The generated new shear strength parameter group is input into the physical model to calculate the corresponding safety factor sample value. The joint reliability evaluation index under the two states is obtained based on the statistical results of all safety factor sample values. The weighted evaluation module determines the weight coefficient of each parameter based on the degree of influence of each shear strength parameter on landslide stability for the natural stable state parameter group and the saturated unstable state parameter group. It also calculates the individual reliability evaluation index for each shear strength parameter and sums the weights with the corresponding individual reliability evaluation indexes to obtain the weighted reliability evaluation indexes for the two states. The decision optimization module compares the joint reliability evaluation index and the weighted reliability evaluation index of the two states with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, the joint reliability evaluation index is selected as the final reliability evaluation index.

[0098] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0099] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0100] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A dynamic assessment method for soil-rock composite landslides based on multi-source data fusion, characterized in that, The specific steps include: Several data collection points were set up in the target landslide area, and the shear strength parameters of each data collection point were obtained under both natural and saturated conditions. A physical model was constructed based on the topography and rock layer distribution characteristics of the landslide. Input the shear strength parameters of each sampling point under the two states into the physical model to obtain the safety factor of each sampling point under the two states. Determine the stable state based on the safety factor. Statistically count the shear strength parameters of the sampling points determined to be stable under the natural state as the natural stable state parameter group, and the shear strength parameters of the sampling points determined to be unstable under the saturated state as the saturated unstable state parameter group. For the parameter sets of natural stable state and saturated unstable state, the probability distribution type of each shear strength parameter in each set is determined. Monte Carlo random sampling is performed according to the probability distribution type. The generated new shear strength parameter set is input into the physical model to calculate the corresponding safety factor sample value. The joint reliability evaluation index under the two states is obtained based on the statistical results of all safety factor sample values. For the parameter groups of natural stable state and saturated unstable state, the weight coefficient of each parameter is determined according to the degree of influence of each shear strength parameter on landslide stability in each group, and the individual reliability evaluation index of each shear strength parameter is calculated. The weight is summed with the corresponding individual reliability evaluation index to obtain the weighted reliability evaluation index for the two states. The joint reliability evaluation index and the weighted reliability evaluation index for the two states are compared with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, the joint reliability evaluation index is selected as the final reliability evaluation index.

2. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 1, characterized in that: Several sampling points were set up in the target landslide area. When obtaining the shear strength parameters of each sampling point under natural and saturated conditions, a typical profile consistent with the main sliding direction was selected in the target landslide area. The typical profile runs through the front, middle and rear edges of the landslide. Several sampling points were set up along the typical profile. Soil and rock samples were obtained at each sampling point under natural and saturated conditions.

3. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 2, characterized in that: Tests were conducted on the collected soil and rock samples to obtain the shear strength parameters of each collection point under natural and saturated conditions. The shear strength parameters included unit weight, cohesion, and internal friction angle.

4. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 3, characterized in that: Based on the topography and rock strata distribution characteristics of the target landslide, a physical model of the target landslide is constructed. The specific steps are as follows: Based on landslide topographic data, a two-dimensional limit equilibrium geometric model for landslide analysis was established. Based on rock layer distribution data, the landslide rock layer sequence was divided from top to bottom into an overlying sliding layer and an underlying non-sliding bedrock layer, and corresponding shear strength parameters were assigned to each rock layer. By setting the material constitutive model and interstrip force functions, setting the model boundary conditions based on topographic data and rock layer distribution data, and setting the potential sliding surface as the sliding surface along the basement interface, a landslide physical model is obtained.

5. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 4, characterized in that: Input the shear strength parameters of each sampling point under both states into the physical model to obtain the safety factor of each sampling point under both states, and compare it with the preset threshold to obtain the parameter set of the natural stable state and the parameter set of the saturated unstable state. The specific steps are as follows: Assign the shear strength parameter of each acquisition point to the element in the physical model corresponding to the spatial location of that acquisition point. The shear strength parameters of each sampling point in its natural state are used as a set of input physical models to calculate the safety factor of the sampling point in its natural state; the shear strength parameters of each sampling point in its saturated state are used as a set of input physical models to calculate the safety factor of the sampling point in its saturated state. The safety factor of each sampling point under natural conditions is compared with the safety threshold. The shear strength parameters corresponding to sampling points with safety factors greater than the safety threshold are classified into the natural steady state parameter group. The safety factor of each sampling point under saturation is compared with the safety threshold. The shear strength parameters corresponding to sampling points with a safety factor less than the safety threshold are classified into the saturated unstable state parameter group.

6. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 3, characterized in that: The joint reliability evaluation index for the natural stable state and the saturated unstable state is obtained through the following steps: The shear strength parameter data were divided into two groups: one group corresponds to the parameter data of the natural stable state, and the other group corresponds to the parameter data of the saturated unstable state. For these two sets of data, the probability distribution types of each parameter, including density, cohesion, and internal friction angle, are identified. The probability distribution types include normal distribution and log-normal distribution. For parameter data in both the naturally stable state and the saturated unstable state, the mean and standard deviation of specific gravity, cohesion, and internal friction angle are calculated respectively. For parameters identified as log-normally distributed, the logarithmic mean and logarithmic standard deviation are calculated after taking the logarithm of the original data. For the natural steady state: Based on the probability distribution type of each parameter under the natural steady state, for parameters that follow a normal distribution, their mean and standard deviation are directly used to generate normal random numbers. For parameters that follow a log-normal distribution, their logarithmic mean and standard deviation are used to generate log-normal random numbers. The Monte Carlo method is used for random sampling to generate N sets of shear strength parameter combinations. Each combination contains a unit weight value, a cohesion value, and an internal friction angle value. For saturated unstable state: Based on the probability distribution type of each parameter under saturated unstable state, for parameters that follow a normal distribution, their mean and standard deviation are used to generate normal random numbers. For parameters that follow a log-normal distribution, their logarithmic mean and standard deviation are used to generate log-normal random numbers. The Monte Carlo method is used for random sampling to generate N sets of shear strength parameter combinations. Each set of combinations includes a unit weight value, a cohesive force value, and an internal friction angle value. The generated N sets of shear strength parameters for the natural steady state are successively substituted into the physical model, and the safety factor corresponding to each set of parameters is calculated to obtain N safety factor values ​​for the natural steady state. The generated N sets of shear strength parameters for the saturated unstable state are successively substituted into the physical model, and the safety factor corresponding to each set of parameters is calculated by the limit equilibrium method to obtain N safety factor values ​​for the saturated unstable state. Based on the N safety factor values ​​under the natural steady state, statistical analysis is performed to calculate the joint reliability evaluation index under the natural steady state. The joint reliability evaluation index includes the average safety factor, the standard deviation of the safety factor, the instability probability, and the reliability index. Based on the N safety factor values ​​under saturated unstable state, statistical analysis is performed to calculate the joint reliability evaluation index under saturated unstable state. The joint reliability evaluation index includes the average safety factor, the standard deviation of the safety factor, the instability probability, and the reliability index.

7. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 6, characterized in that: The weighted reliability evaluation index for the naturally stable state and the saturated unstable state is obtained through the following steps: For both the naturally stable state and the saturated unstable state, the average value of each parameter is used as the baseline value. Within a 100% range, the value of one parameter is changed successively while keeping the values ​​of other parameters unchanged. The combination of parameters after each change is substituted into the physical model to calculate the corresponding safety factor. The sensitivity change curves of each parameter are plotted based on the rate of change of the safety factor. The degree of influence of each parameter on landslide stability is determined by the absolute value of the slope of the curve. Based on the degree of influence of each parameter, the weighting coefficients of specific gravity, cohesion and internal friction angle are calculated using an improved three-scale analytic hierarchy process. For density, cohesion, and internal friction angle, based on the probability distribution type of the parameter under the corresponding state and the mean and standard deviation of the parameter under the corresponding state, the Monte Carlo method is used to randomly sample and generate multiple sets of random values ​​for the parameter. Each set of random values ​​is substituted into the physical model in turn to calculate the safety factor. At the same time, the other two parameters are fixed as the average value. The single reliability index of the parameter is obtained by statistical analysis based on the calculated safety factor value. The weighted reliability index is obtained by multiplying the weighting coefficients of each parameter by the individual reliability index of each parameter and then summing the results. Repeat the above steps for both the naturally stable state and the saturated unstable state to obtain the weighted reliability index for the naturally stable state and the weighted reliability index for the saturated unstable state.

8. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 7, characterized in that: The weighting coefficients for specific gravity, cohesion, and angle of internal friction are calculated using an improved three-scale analytic hierarchy process. The specific steps are as follows: Based on the ranking of the influence of density, cohesion, and internal friction angle on landslide stability under natural conditions, a three-scale judgment matrix A is constructed, where the matrix contains the... Line 1 Column elements Values ​​are determined according to the following rules: When the The parameter is compared to the first When a parameter is important The value is 1; When the The parameter is compared to the first When both are equally important, The value is 0; When the The parameter is compared to the first When a parameter is important The value is -1; Based on the three-scale judgment matrix A, calculate the optimal transfer matrix B, where the matrix contains the first... Line 1 Column elements Equal to the first in matrix A Line 1 The average of the sums of corresponding elements in a column; Based on the optimal transfer matrix B, calculate the judgment matrix C, where the first... No. Column elements Equal to the natural constant e as the base, The power of the exponent; Based on the judgment matrix C, calculate the set average of all elements in each row to obtain the initial weights of each parameter. Then normalize each initial weight so that the sum of the weight coefficients of the three parameters equals 1, and obtain the final weight coefficients of specific gravity, cohesion and internal friction angle. Based on the degree of influence of unit weight, cohesion, and internal friction angle on landslide stability under saturated unstable conditions, a three-scale judgment matrix A is constructed. The matrix contains the first... Line 1 Column elements Values ​​are determined according to the following rules: When the The parameter is compared to the first When a parameter is important The value is 1; When the The parameter is compared to the first When both are equally important, The value is 0; When the The parameter is compared to the first When a parameter is important The value is -1; Based on the three-scale judgment matrix A, calculate the optimal transfer matrix B, where the matrix contains the first... Line 1 Column elements Equal to the first in matrix A Line 1 The average of the sums of corresponding elements in a column; Based on the optimal transfer matrix B, calculate the judgment matrix C, where the first... No. Column elements Equal to the natural constant e as the base, The power of the exponent; Based on the judgment matrix C, calculate the set average of all elements in each row to obtain the initial weights of each parameter. Then normalize each initial weight so that the sum of the weight coefficients of the three parameters equals 1, thus obtaining the final weight coefficients of density, cohesion and internal friction angle.

9. The dynamic assessment method for rock and soil composite landslides based on multi-source data fusion according to claim 1, characterized in that: The joint reliability evaluation index and weighted reliability evaluation index for the two states are compared with preset accuracy thresholds and efficiency thresholds, respectively. The specific steps are as follows: Using the reliability index in the weighted reliability evaluation index as the benchmark value, the accuracy threshold is set to k times the reliability index, where k is a preset coefficient greater than 1. Using the calculation time of the joint reliability evaluation index as the benchmark value, the efficiency threshold is set as 1 / m of the joint reliability evaluation index, where m is a preset coefficient greater than 1. The reliability index in the joint reliability evaluation index is compared with the accuracy threshold. If the reliability index reaches or exceeds the accuracy threshold, the joint reliability evaluation index is determined to meet the accuracy threshold requirement. The actual calculation time of the weighted reliability evaluation index is compared with the efficiency threshold. If the calculation time is less than or equal to the efficiency threshold, the weighted reliability evaluation index is determined to meet the efficiency threshold requirement. If the joint reliability assessment index meets the accuracy threshold requirement, then the joint reliability assessment index is selected as the final reliability assessment index. If the joint reliability assessment index does not meet the accuracy threshold requirement, but the weighted reliability assessment index meets the efficiency threshold requirement, then the weighted reliability assessment index shall be selected as the final reliability assessment index. If the joint reliability assessment index does not meet the accuracy threshold requirement, and the weighted reliability assessment index does not meet the efficiency threshold requirement, then the weighted reliability assessment index shall be selected as the final reliability assessment index.

10. A dynamic assessment system for rock and soil composite landslides based on multi-source data fusion, characterized in that: The evaluation system is used to execute the dynamic evaluation method for rock and soil composite landslides based on multi-source data fusion as described in any one of claims 1-9, including: The model building module sets up several collection points in the target landslide area and obtains the shear strength parameters of each collection point under natural and saturated conditions. It then constructs a physical model based on the landslide's topography and rock layer distribution characteristics. The parameter grouping module inputs the shear strength parameters of each sampling point under two states into the physical model to obtain the safety factor of each sampling point under the two states. Based on the safety factor, the stable state is determined. The shear strength parameters of the sampling points that are determined to be stable under the natural state are counted as the natural stable state parameter group, and the shear strength parameters of the sampling points that are determined to be unstable under the saturated state are counted as the saturated unstable state parameter group. The joint evaluation module determines the probability distribution type of each shear strength parameter in each of the natural stable state parameter group and the saturated unstable state parameter group. Monte Carlo random sampling is performed according to the probability distribution type. The generated new shear strength parameter group is input into the physical model to calculate the corresponding safety factor sample value. The joint reliability evaluation index under the two states is obtained based on the statistical results of all safety factor sample values. The weighted evaluation module determines the weight coefficient of each parameter based on the degree of influence of each shear strength parameter on landslide stability for the natural stable state parameter group and the saturated unstable state parameter group. It also calculates the individual reliability evaluation index for each shear strength parameter and sums the weights with the corresponding individual reliability evaluation indexes to obtain the weighted reliability evaluation indexes for the two states. The decision optimization module compares the joint reliability evaluation index and the weighted reliability evaluation index of the two states with the preset accuracy threshold and efficiency threshold, respectively. When the joint reliability evaluation index meets the accuracy threshold requirement, the joint reliability evaluation index is selected as the final reliability evaluation index.

Images