A rainfall-flood coupling sliding probability prediction method based on physical model test

Abstract

The application discloses a rainfall-flood coupling sliding probability prediction method based on physical model test, and the method is based on physical model test and multi-source monitoring data to construct a standardized multi-dimensional time sequence input matrix. Firstly, the effective normal stress sequence of the sliding zone is calculated in real time, the time-varying strength parameters are dynamically inverted through the nonlinear reduction model of the rock-soil body strength, and the geometric topology of the landslide is updated in real time. The improved limit equilibrium state equation considering the dynamic water pressure and the change of the geometric shape is established, and the deterministic safety factor at any time is solved. The probability distribution of the key mechanical and hydraulic parameters is determined, a plurality of parameter samples are generated by combining the Monte Carlo simulation random sampling, and the improved limit equilibrium equation is substituted to calculate the safety factor sample sequence in batches. Finally, the frequency of the values lower than the critical threshold is counted, the landslide instability probability at the current time is obtained, and the dynamic evaluation of the landslide stability under the rainfall-flood double action from the probability is realized.

Description

Technical Field

[0001] This invention belongs to the field of landslide model testing technology, specifically relating to a method for predicting the probability of landslides caused by rainfall-flood coupling based on physical model testing. Background Technology

[0002] A landslide is a natural phenomenon in which rock and soil on a slope slides down the slope as a whole or in parts under the influence of gravity, influenced by factors such as river erosion, groundwater activity, rainwater soaking, earthquakes, and artificial slope cutting.

[0003] Traction landslides are a common type of slope instability with specific development patterns. Their core characteristic is that the toe of the slope first fails and then pulls the upper rock and soil mass down step by step, resulting in overall or local tensile failure. This type of traction landslide not only directly threatens the lives of residents below and around the slope, but also causes serious damage to various infrastructures such as houses, roads, and water conservancy facilities, resulting in huge property losses. At the same time, the sliding of rock and soil mass generated by the landslide will also destroy the vegetation cover of the surrounding area, causing problems such as soil erosion and desertification, and causing long-term and difficult-to-repair negative impacts on the local natural ecological environment.

[0004] Among the inducing factors of traction landslides, the failure at the toe of the slope is often the key initiating link that triggers the entire landslide disaster. Among the many factors that lead to the failure of the toe of the slope, the synergistic effect of rainfall infiltration into the slope and river erosion of the toe of the slope has a decisive influence on the instability of riverbank slopes or slopes near water bodies.

[0005] Currently, in the field of landslide disaster research, relatively mature landslide risk prediction methods have been established. However, for traction landslides caused by the combined effects of rainfall infiltration and flood erosion, the prediction effect is poor due to the poor geological conditions of traction landslides, the difficulty of on-site monitoring and construction, the high cost of data collection, and the poor prediction effect. At present, there is no rainfall-flood coupled landslide probability prediction method that can simultaneously meet the requirements of simple operation, high practical application value, and accurate and effective prediction effect.

[0006] Therefore, inventing a method that can accurately simulate the combined effects of rainfall and flood and predict the probability of landslides caused by rainfall-flood coupling has become an urgent need in the field of landslide disaster research. Summary of the Invention

[0007] This invention provides a method for predicting the probability of landslides caused by rainfall-flood coupling based on physical model tests. It aims to improve the accuracy and precision of landslide risk prediction under the dual effects of rainfall infiltration and flood scouring by simulating the failure phenomenon of traction landslides caused by the combined effects of rainfall infiltration and flood scouring. It achieves a progressive dynamic assessment of landslide stability from deterministic to probabilistic, while the test conditions are easier to meet. It has the advantages of strong simulation capability, ease of use, and low economic cost.

[0008] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0009] In a first aspect, the present invention provides a method for predicting the probability of landslides caused by rainfall-flood coupling based on physical model experiments, comprising the following steps;

[0010] Step S1: Construct a physical model test system to simulate the landslide environment to be evaluated, and conduct simulation tests based on the physical model test system. Acquire and preprocess multi-source monitoring data during the test process. The multi-source monitoring data includes raw data of internal physical field, external morphological field and boundary hydraulic field. Construct a standardized multi-dimensional time series input matrix.

[0011] Step S2: Calculate the effective normal stress sequence at the sliding section in real time based on the input matrix;

[0012] Step S3: Construct a nonlinear reduction model for the strength of the soil and rock mass, and dynamically invert the time-varying strength parameters that change with time;

[0013] Step S4: Update the geometric topology of the landslide body in real time to obtain the updated strip geometric parameters;

[0014] Step S5: Based on the effective normal stress sequence, time-varying intensity parameters, and updated strip geometric parameters at the sliding zone, the deterministic safety factor at any time is calculated according to the improved limit equilibrium state equation considering the action of hydrodynamic pressure and changes in geometric shape.

[0015] Step S6: Analyze the spatial variability and measurement error of material parameters, and determine the probability distribution of key mechanical and hydraulic parameters;

[0016] Step S7: Set the number of Monte Carlo simulations, and randomly sample based on the probability distribution of the determined key mechanical and hydraulic parameters to generate multiple sets of random parameter sample sets;

[0017] Step S8: Substitute each set of random parameter samples into the improved limit equilibrium state equation to calculate the corresponding safety factor sample sequence in batches;

[0018] Step S9: Count the frequency of the safety factor sample sequence that is less than the critical stability threshold, and calculate the landslide instability probability value at the current moment.

[0019] Furthermore, in step S1, the construction process of the standardized multidimensional time-series input matrix includes:

[0020] Acquire internal physics data sequence ,in, for The pore water pressure measured by the time sensor. for Total earth pressure measured by the sensor at any given time. for Soil saturation measured by a time sensor;

[0021] Acquire external morphological field data sequence ,in, for The depth of the scour pit at the foot of the slope at any given moment;

[0022] Obtain boundary hydraulic field data sequence ,in, for The flow rate at any given moment. for The water level at any given time;

[0023] A sliding window filtering algorithm is used to remove high-frequency noise, and linear interpolation alignment is performed based on millisecond-level timestamps to form a standardized multidimensional time-series input matrix. :

[0024] ;

[0025] In this context, the superscript T indicates matrix transpose.

[0026] Furthermore, in step S2, the formula for calculating the effective normal stress sequence is:

[0027]

[0028] In the formula: express The effective normal stress at the constant sliding section;

[0029] express Total earth pressure measured by the time sensor;

[0030] express Pore ​​water pressure measured by a time sensor;

[0031] This represents the effective stress coefficient, which is taken as 1.0 for saturated soil and according to... Linear interpolation is used to determine this.

[0032] Furthermore, in step S3, the time-varying intensity parameters include effective cohesion and effective internal friction angle;

[0033] The specific expression for the nonlinear reduction model of soil and rock strength is as follows:

[0034]

[0035]

[0036] In the formula: express Effective cohesion at any given time;

[0037] express Effective internal friction angle at any given time;

[0038] These represent the initial effective cohesion and initial effective internal friction angle of the model material, respectively.

[0039] express The soil saturation at time t, with a value range of [0,1].

[0040] express The depth of scour at the toe of the slope at any given moment;

[0041] This indicates the initial height of the bank slope model;

[0042] Represents the natural exponential function; All of these represent the material damage coefficient, calibrated by pre-testing; This represents the nonlinear exponent, typically ranging from 1.0 to 2.0.

[0043] Furthermore, in step S4, the geometric topology of the landslide body is updated in real time to obtain the updated strip geometric parameters, including:

[0044] Reconstructing 3D point cloud data using a 3D laser scanner Slope surface equation at time Identify the scour boundary at the toe of the slope;

[0045] The landslide mass is divided along the potential sliding surface into... The nth vertical strip, for the nth Individual blocks:

[0046] Update # The bottom arc length of each strip : In the initial state, the first The bottom arc segment of each strip Based on the time... Reconstructing the scour airspace at the foot of the slope from 3D point cloud data and the initial slope area After deducting the scour void, the current remaining soil area is obtained:

[0047]

[0048] The first The intersection of the initial bottom slip arc segment of each block with the current remaining soil area yields the effective slip arc segment that is still in contact with the soil:

[0049]

[0050] No. The updated bottom arc length for each strip is:

[0051]

[0052] When the airspace is washed away and the first When the bottom sliding arc segments of individual blocks overlap, the sliding arc segment corresponding to the overlapping part is determined to be an exposed segment or a suspended segment, and is no longer included in the effective anti-slip arc length.

[0053] In the formula, This refers to the slope body in its initial state. For a moment The slope toe scour airspace identified based on 3D point cloud data; For a moment The current remaining soil area after deducting the scoured airspace; For the initial state, the first The bottom arc segment of each strip; For a moment No. The effective bottom slip arc segment that is still in contact with the remaining soil area; For a moment No. The updated effective bottom arc length of each strip block; Operator for calculating curve length; For monitoring time; Number the blocks. , This represents the total number of blocks.

[0054] when and When overlap exists, the slip arc segment corresponding to the overlapping portion is determined to be either an exposed scour segment or a suspended segment, and is no longer included in the calculation. ;when Completely located At that time, Equal to the initial bottom arc length; when Completely fall into At that time, .

[0055] The first Volume of each strip The update method is as follows:

[0056]

[0057] in, For the first Each block at time The horizontal projection area The elevation of the current slope surface is obtained from the reconstruction of the 3D point cloud. For potential slip surface elevation, This means that if the value inside the parentheses is greater than 0, the original value is used; otherwise, 0 is used.

[0058] Update # Each block's weight The calculation formula is:

[0059]

[0060] In the formula: express Saturation intensity at any given moment; Indicates natural severe; Indicates the first The volume of each strip changes dynamically with the scouring geometry; Indicates the first Soil saturation of each block.

[0061] Furthermore, in step S5, based on the effective normal stress sequence, time-varying intensity parameters, and updated strip geometric parameters at the sliding zone, and according to the constructed improved limit equilibrium state equation considering hydrodynamic pressure and geometric changes, the deterministic safety factor at any given time is calculated, including:

[0062] Calculate the first The anti-slip moment of each strip is:

[0063]

[0064] Among them, the Effective force of each block normal direction The following was calculated based on the effective normal stress sequence at the slip zone:

[0065]

[0066] In the formula, The effective normal stress sequence at the slip zone The first obtained through spatial mapping or interpolation The average effective normal stress at the bottom of each strip, specifically, when the effective normal stress of the sliding strip is... When multiple monitoring points provide the data, the average effective normal stress is expressed as: In the formula, For the first Each slip zone monitoring point at time The effective normal stress; For the first The monitoring point for the first Spatial interpolation weights for each block, and satisfying .

[0067] For the first Effective sliding arc length at the bottom of each strip block For the first The width of each strip is calculated (under the condition of calculating the width per unit area, take...). ).

[0068] Calculate the first The downward torque of each block :

[0069] ;

[0070] Among them, the first The dynamic water pressure of each block The calculation is as follows:

[0071] ;

[0072] Calculate the deterministic safety factor:

[0073] ;

[0074] In the formula: : radius of the arc : No. Inclination angle of the strip surface; : Coefficient of drag of an object : Water density, n is the total number of strips; : No. The projected area of ​​the strip in the direction of water flow changes dynamically with scouring. : The angle between the water flow vector and the tangent of the slip surface; This represents the component of the seepage force in the normal direction.

[0075] Furthermore, in step S6, the spatial variability of material parameters and measurement errors are analyzed to determine the probability distribution of key mechanical and hydraulic parameters, including:

[0076] Effective cohesion internal friction angle Near-bottom flow velocity The probability distribution type is determined to be either normal or log-normal, and its statistical characteristic values ​​are determined as follows:

[0077] mean Instantaneous values ​​of effective cohesion, internal friction angle, and near-bottom velocity derived from inversion; standard deviation Based on the coefficient of variation of the material homogeneity test Sure: ;

[0078] The coefficient of variation of cohesion The coefficient of variation of the internal friction angle is taken as 0.15~0.25. The coefficient of variation for flow velocity measurements is taken as 0.05~0.10. Take a value of 0.03~0.05;

[0079] Define the depth of the scour pit at the toe of the slope. The depth of the scour pit at the toe of the slope at this moment is randomly distributed within the scanning error range.

[0080] Furthermore, in step S7, the number of Monte Carlo simulations is set, and random sampling is performed based on the probability distributions of the determined key mechanical and hydraulic parameters to generate multiple sets of random parameter sample sets, including:

[0081] Set the total number of simulations ;

[0082] Use the Mersenne Twister algorithm to generate a pseudo-random number sequence;

[0083] For the j-th simulation:

[0084] Generate a random number seed and draw the first random number from the defined distribution. Group of random parameter vectors ;

[0085] in, This is the effective cohesion sample for the j-th simulation; This is the sample of the internal friction angle in the j-th simulation; This is the near-bottom velocity sample from the j-th simulation; This is a sample of the slope toe scour pit depth from the j-th simulation.

[0086] Furthermore, in step S8, each set of random parameter samples is substituted into the improved limit equilibrium state equation to calculate the corresponding safety factor sample sequence in batches, including:

[0087] The first Group of random parameter vectors Substituting the parameters into the formula for calculating the deterministic safety factor, while keeping the other parameters unchanged, we can calculate the first... Safety factor of this simulation Finally, a safety factor sample set is obtained. .

[0088] Furthermore, in step S9, the frequency of values ​​in the safety factor sample sequence that are less than the critical stability threshold is counted, and the landslide instability probability value at the current moment is calculated, including:

[0089] landslide instability probability value The calculation formula is:

[0090] ;

[0091] In the formula: This is an indicator function; it takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. This is the critical stability threshold; The total number of simulations for Monte Carlo.

[0092] Furthermore, the method also includes:

[0093] Based on the landslide instability probability value and the deterministic safety factor, a preset graded early warning rule is matched, and multi-level early warning signals are output.

[0094] The tiered early warning rules include:

[0095] like and It outputs a first-level warning signal to indicate a safe state;

[0096] like or It outputs a level-two early warning signal to alert users and increases the monitoring frequency.

[0097] like or It outputs a three-level early warning signal to indicate the risk of local shear damage;

[0098] like or It outputs a level four early warning signal to alert that a landslide is about to become unstable, and simultaneously outputs the remaining stabilization time extrapolated from the current trend.

[0099] Secondly, the present invention provides a rainfall-flood physical model test system, comprising:

[0100] A bank slope model box is used to build and contain a soil and rock mass model simulating a bank slope.

[0101] A rainfall system, erected above the bank slope model box, is used to apply artificial rainfall;

[0102] The river flow system is connected to one side of the bank slope model box and is used to simulate the river flow or reservoir water level environment.

[0103] The integrated monitoring system includes multiple sensors distributed inside the bank slope model box and in the river flow system, used to collect multi-source monitoring data during the experiment. The multi-source monitoring data includes raw data of the internal physical field, external morphological field and boundary hydraulic field.

[0104] A data processing terminal, connected to the integrated monitoring system, has a built-in processor and memory. The processor is configured to execute the method described in the first aspect to output a landslide instability probability prediction result.

[0105] The integrated monitoring system includes:

[0106] The internal physical field monitoring unit includes pore water pressure sensors, earth pressure cells, frequency domain reflectometers, and deep displacement gauges embedded in the slope toe and within the slope model, used to acquire raw data of the internal physical field.

[0107] The external morphology monitoring unit includes a high-speed camera and a 3D laser scanner deployed on the outside of the bank slope model box and the river flow system, which are used to acquire raw data of the external morphology field in real time.

[0108] The boundary hydraulic field monitoring unit includes a flow meter, a water level gauge, and a tipping bucket rain gauge installed in the river flow system to acquire raw boundary hydraulic field data in real time.

[0109] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0110] 1. This method constructs a scaled-down, isomaterial model of a landslide site under the combined effects of rainfall infiltration and flood erosion indoors. Multiple parameter samples are generated through random sampling using Monte Carlo simulation, and these samples are substituted into the improved limit equilibrium equation to calculate a batch of safety factor sample sequences. Finally, the frequency of landslides falling below the critical threshold is statistically analyzed to obtain the probability of landslide instability at the current moment. This achieves a progressive dynamic assessment of landslide stability from deterministic to probabilistic, effectively improving the accuracy and precision of landslide risk prediction. Furthermore, compared to large-scale in-situ monitoring of landslides under the combined effects of rainfall infiltration and flood erosion, this method offers advantages such as relatively controllable material costs, reusable models, consistent initial conditions for each test, ensuring the feasibility of repeated tests and the statistical reliability of results. The experimental conditions are easier to meet, and this method boasts advantages such as strong simulation capabilities, ease of use, and low economic cost.

[0111] 2. This method solves the challenge of landslide field testing under the dual effects of rainfall infiltration and flood erosion, ensuring accurate and reliable results. The scaled-down, same-material model avoids irreversible environmental damage and safety risks associated with large-scale landslide field testing. It also ensures that the physical and mechanical properties of the soil and rock are strictly consistent with the prototype, overcoming the size effect of scaled-down models. This allows test results to directly reflect deformation and failure patterns at a real scale, eliminating the need for conversion and significantly improving the authenticity and reliability of the test.

[0112] 3. Supports repeated tests and comparisons under multiple conditions, improving efficiency and accuracy. This system can simulate the impact of single factors (such as rainfall, floods, etc.) or the coupled effects of multiple factors on landslide stability under controlled environments, enabling slope stability studies under continuously changing parameters and significantly saving material and time costs. Combined with transparent test chambers or high-precision monitoring equipment, the entire process of a landslide from stability to instability can be visually presented.

[0113] 4. Acquire internal response data in multiple dimensions and throughout the entire lifecycle to fill gaps in on-site monitoring. By pre-embedding sensors during model building, "holographic perception" of the internal mechanical behavior of the landslide body can be achieved. Furthermore, with the help of devices such as 3D laser scanning and displacement gauges, the evolution process of deep displacement and plastic zone can be obtained, fully reproducing the gradual failure of the slip zone and the entire process of landslide from creep to abrupt change, providing a reliable data foundation for risk assessment.

[0114] 5. Providing physical data for landslide probability prediction and graded early warning. This scheme uses physical simulation to obtain the physical laws and critical thresholds of landslide instability under different inducing factors, supporting the establishment of a physical-driven probabilistic judgment model. Based on data such as slip zone stress-strain curves and pore water pressure, refined graded early warning criteria can be established.

[0115] 6. Multi-hazard-causing factor coupled simulation was realized: By organically integrating the three subsystems of bank slope simulation, rainfall simulation and river flow simulation, the real, synchronous or sequential simulation of the key coupled hazard-causing factor of "rainfall infiltration" and "flood scouring (or reservoir water action)" was realized for the first time on a physical platform, which can more realistically reproduce the triggering conditions and evolution environment of natural landslides.

[0116] 7. A comprehensive three-dimensional monitoring network was constructed: The integrated monitoring system combines contact and non-contact sensing technologies to achieve synchronous, three-dimensional, and multi-dimensional data acquisition of the internal physical field, external morphological field, and boundary hydraulic field during landslide evolution. This provides unprecedented and complete data chain support for deeply revealing the spatiotemporal linkage mechanism and chain generation law between internal softening of the slope and external erosion at the slope toe under the coupling effect of rainfall and flood.

[0117] 8. Enhanced the flexibility and scientific rigor of model testing: Key parameters such as bank slope, riverbed slope, rainfall intensity, and water flow velocity / level can be independently and precisely controlled, enabling a single system to simulate various geological and topographical conditions and disaster scenarios of varying intensities, greatly expanding the scope and depth of experimental research. Layered filling and material proportioning techniques improved the similarity between the physical model and the prototype. Attached Figure Description

[0118] Figure 1 This is a schematic diagram of the test system of the present invention;

[0119] Figure 2 This is a flowchart of the rainfall-flood coupled landslide probability prediction method based on physical model experiments according to the present invention.

[0120] In the diagram: 1. Bank slope model box; 2. Support frame; 3. Main water supply pipe; 4. Branch pipe; 5. Atomizing nozzle; 6. Hand-cranked screw jack; 7. Water storage tank; 8. Variable frequency water pump; 9. Water pipe; 10. Control valve; 11. Rectifier grid; 12. Water tank; 13. Baffle plate; 14. Base. Detailed Implementation

[0121] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0122] In the description of this embodiment, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this embodiment and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this embodiment.

[0123] Example 1:

[0124] This embodiment provides a method for predicting the probability of landslides caused by coupled rainfall and flood based on physical model experiments. This method relies on a "physical model experiment system" (which includes a bank slope model box, a rainfall system, a river flow system, and a comprehensive monitoring system; for specific hardware structure, please refer to the accompanying drawings and embodiment descriptions). Specifically, the physical simulation experiment system is constructed based on the landslide environment to be evaluated, using materials of the same scale, and can simulate extreme situations, such as... Figure 1 As shown, the specific structure includes:

[0125] Slope Model Box 1 is used to build and contain the soil and rock mass model simulating the slope;

[0126] The rainfall system is installed above the slope model box 1 and is used to apply artificial rainfall; the rainfall amount is set proportionally according to the predicted maximum rainfall in the landslide environment and the model construction ratio.

[0127] The river flow system is connected to one side of the bank slope model box 1 and is used to simulate the river flow or reservoir water level environment; the flow rate is set proportionally according to the predicted maximum flow rate and the model construction ratio.

[0128] The integrated monitoring system includes multiple sensors distributed inside the bank slope model box 1 and in the river flow system, which are used to synchronously collect multi-physics parameters;

[0129] The integrated monitoring system includes:

[0130] The internal physical field monitoring unit includes pore water pressure sensors, earth pressure cells, frequency domain reflectometers and deep displacement gauges embedded in the slope toe and the slope body, used to acquire raw data of the internal physical field;

[0131] The external morphology monitoring unit includes a high-speed camera and a 3D laser scanner deployed on the outside of the bank slope model box and the river flow system, which are used to acquire raw data of the external morphology field in real time.

[0132] The boundary hydraulic field monitoring unit includes a flow meter, a water level gauge, and a tipping bucket rain gauge installed in the river flow system to acquire raw boundary hydraulic field data in real time.

[0133] The core innovation of this invention lies in establishing a complete calculation process from multi-source monitoring data to probabilistic risk criteria. This method overcomes the shortcomings of traditional physical simulation experiments, which can only qualitatively observe the failure mode and cannot quantitatively assess the risk of transient instability.

[0134] like Figure 2 As shown, the specific implementation steps of this method are as follows:

[0135] Step S1: Synchronous acquisition and standardized input of multi-source monitoring data. The data source for this step is the integrated monitoring system, specifically including three types of data streams:

[0136] (1) Internal physical field data :

[0137] : Pore water pressure at time t, reading of pore water pressure sensor (kPa) buried at the potential slip zone.

[0138] : Total earth pressure at time t, earth pressure cell reading (kPa) at the potential slip zone.

[0139] : Soil saturation at time t, saturation converted from volumetric water content measured by frequency domain reflectometer (TDR) (dimensionless).

[0140] : The horizontal displacement of the deep layer at time t, the horizontal displacement (mm) measured by the deep displacement gauge.

[0141] (2) External morphological field data :

[0142] : The depth of the scour pit at the toe of the slope at time t, the maximum depth of the scour pit at the toe of the slope calculated by the point cloud of the 3D laser scanner (m).

[0143] : The width of the slope surface crack at time t, the average width (mm) of the slope surface tension crack extracted by the image recognition algorithm.

[0144] : Surface displacement vector at time t, surface displacement vector field (mm) calculated by digital image correlation (DIC) technique.

[0145] (3) Boundary hydraulic field data :

[0146] : The incoming flow velocity at time t, the near-bottom velocity (m / s) measured by an acoustic Doppler current meter (ADV).

[0147] : Water level at time t, water level measured by ultrasonic level gauge (m).

[0148] : Rainfall intensity at time t, the rainfall intensity (mm / h) recorded by the tipping bucket rain gauge.

[0149] Because the sampling frequencies of different sensors are inconsistent (e.g., displacement gauge 1Hz, laser scanner 0.1Hz, pressure sensor 10Hz), a unified timestamp is first applied to all data using a high-precision clock source. Then, a sliding window Savitzky-Golay filtering algorithm (window width 0.5s, polynomial order 2) is used to remove high-frequency electronic noise and mechanical vibration interference. Finally, cubic spline interpolation is used to resample all data to a unified time step. Construct a standardized input matrix:

[0150]

[0151] in For the first Each time step For the first A standardized multidimensional time-series input matrix for each time step.

[0152] This step ultimately yields a standardized, denoised, and time-aligned multidimensional time series matrix. This serves as the foundational data for all subsequent calculations. By constructing a physical model test system and integrating the original data from the internal physical field, external morphological field, and boundary hydraulic field, a standardized multidimensional time-series input matrix was established. This effectively solved the problem of the difficulty in synchronously utilizing heterogeneous data in landslide monitoring, providing a high-quality, time-aligned input foundation for subsequent dynamic analysis.

[0153] Step S2: Real-time calculation of effective normal stress based on the effective stress principle, according to the results obtained in step S1. and Based on the effective stress principle of soil mechanics and considering the characteristics of unsaturated soil, the effective normal stress at the slip zone is calculated. The calculation formula is as follows:

[0154]

[0155] The effective normal stress (kPa) at a given time is a key stress indicator that determines the shear strength of soil.

[0156] : Total stress (kPa) measured by the time sensor.

[0157] : The pore water pressure (kPa) is measured at all times. Positive values ​​represent pressure, and negative values ​​represent suction.

[0158] : Effective stress coefficient. When hour, ;when hour, This embodiment uses a linear approximation: .

[0159] Step S3: Dynamic inversion of soil and rock mechanical parameters and strength reduction, based on the monitoring data from step S1. , And the initial parameters of the model material: true effective cohesion. and internal friction angle Considering the soil softening (loss of matrix suction) caused by rainfall infiltration and the disturbance of the soil structure at the slope toe caused by flood erosion, a two-factor coupled strength reduction model is constructed.

[0160] The effective cohesion time-varying model is as follows:

[0161]

[0162] Time-varying model of effective internal friction angle:

[0163]

[0164] : The actual effective cohesion (kPa) and internal friction angle (°) at any given time.

[0165] Initial strength parameters obtained by direct shear test before the experiment.

[0166] Soil saturation reflects the degree of rainfall infiltration.

[0167] Relative scour depth reflects the proportion by which floods weaken the slope's support capacity at the toe.

[0168] Material damage coefficient. For example, for a silty clay model, .

[0169] : Nonlinear exponent, characterizing the softening rate, usually taken as 1.5. This represents the natural exponential function.

[0170] This step implements a dynamically updated sequence of intensity parameters. This truly reflects the deterioration process of soil properties during the experiment.

[0171] This method calculates the effective normal stress sequence at the slip zone in real time and, combined with a nonlinear reduction model of soil and rock strength, dynamically inverts the time-varying strength parameters. This design can realistically reflect the strength degradation process of the landslide slip zone under the influence of water content changes and shear rate, avoiding the errors caused by traditional constant parameter assumptions.

[0172] Step S4: Based on the geometric shape and load correction of the 3D reconstruction, according to the monitoring data of step S1 , and 3D point cloud data, utilizing Reconstructing the slope surface equation from 3D point cloud data at time step The landslide body is divided along the pre-defined sliding surface into... The vertical strips are defined as follows: For the strips near the toe of the slope, the scour space is identified based on 3D point cloud data, the intersection of the initial slip arc segment and the remaining soil area is calculated, and the effective contact arc length is calculated to update the bottom slip arc length in real time. .

[0173] Updating the bottom arc length of the i-th block includes: the bottom arc segment of the i-th block in the initial state. Based on the time... Reconstructing the scour airspace at the foot of the slope from 3D point cloud data and the initial slope area After deducting the scour void, the current remaining soil area is obtained:

[0174]

[0175] Intersecting the initial bottom slip arc segment of the i-th block with the current remaining soil region, we obtain the effective slip arc segment that is still in contact with the soil:

[0176]

[0177] The updated bottom arc length of the i-th block is:

[0178]

[0179] When the scoured airspace overlaps with the bottom sliding arc segment of the i-th block, the sliding arc segment corresponding to the overlapping part is determined to be an exposed segment or a suspended segment, and is no longer included in the effective anti-slip arc length.

[0180] In the formula, This refers to the slope body in its initial state. For a moment The slope toe scour airspace identified based on 3D point cloud data; For a moment The current remaining soil area after deducting the scoured airspace; For the initial state, the first The bottom arc segment of each strip; For a moment No. The effective bottom slip arc segment that is still in contact with the remaining soil area; For a moment No. The updated effective bottom arc length of each strip block; Operator for calculating curve length; For monitoring time; Number the blocks. , This represents the total number of blocks.

[0181] No. Volume of each strip The update method is as follows:

[0182]

[0183] in, For the first Each block at time The horizontal projection area The elevation of the current slope surface is obtained from the reconstruction of the 3D point cloud. For potential slip surface elevation, This means that if the value inside the parentheses is greater than 0, the original value is used; otherwise, 0 is used.

[0184] If the scour depth exceeds the bottom of the strip, the slip length of that section is set to zero (i.e., suspended). Simultaneously, considering the increase in soil weight due to rainwater infiltration, the [number]th section is updated. The weight of each block :

[0185]

[0186] in The volume of the strip changes with scouring.

[0187] This step yields the updated strip geometry parameters. By updating the geometric topology of the landslide body in real time and substituting the updated segment geometric parameters, effective normal stress of the slip zone, and time-varying strength parameters into the improved limit equilibrium equation, while also introducing the effect of hydrodynamic pressure, this method is more adaptable to dynamic changes in landslide morphology and fluctuations in hydraulic conditions compared to the traditional limit equilibrium method, significantly improving the calculation accuracy of the deterministic safety factor.

[0188] Step S5: Construct an improved limit equilibrium equation considering hydrodynamic pressure, based on step S2. Step S3 Geometric parameters of step S4 , and step S1 An improved Swedish slice method is employed, incorporating flood hydrodynamic pressure and seepage force terms. This leads to the determination of the deterministic safety factor. .

[0189] Calculate the first The anti-slip moment of each strip is:

[0190]

[0191] Among them, the Effective force of each block normal direction Calculated based on the effective normal stress sequence of the sliding band:

[0192]

[0193] In the formula, For the effective normal stress sequence of the slip band The first obtained through spatial mapping or interpolation The average effective normal stress at the bottom of each strip, specifically, when the effective normal stress of the sliding strip is... When multiple monitoring points provide the data, the average effective normal stress is expressed as: In the formula, For the first Each slip zone monitoring point at time The effective normal stress; For the first The monitoring point for the first Spatial interpolation weights for each block, and satisfying .

[0194] For the first Effective sliding arc length at the bottom of each strip block For the first The width of each strip is calculated (under the condition of calculating the width per unit area, take...). ).

[0195] Calculate the first The downward torque of each block :

[0196] ;

[0197] Among them, the first The dynamic water pressure of each block The calculation is as follows:

[0198] ;

[0199] Calculate the deterministic safety factor:

[0200] ;

[0201] In the formula: : radius of the arc : No. Inclination angle of the strip surface; : Coefficient of drag of an object : Water density, n is the total number of strips; : No. The projected area of ​​the strip in the direction of water flow changes dynamically with scouring. : The angle between the water flow vector and the tangent of the sliding surface.

[0202] Step S6: Uncertainty analysis and probability distribution modeling, based on the inversion from Step S3. Mean, step S1 The mean is processed. The probability distribution type of the key mechanical and hydraulic parameters is determined to be either normal or log-normal distribution, and the statistical characteristic values ​​are determined as follows:

[0203] mean The instantaneous value is directly taken from the inversion in step S2. and the actual measurement in step S1 Standard deviation Based on the coefficient of variation of the material homogeneity test Determined, that is The coefficient of variation of cohesion The coefficient of variation of the internal friction angle is taken as 0.15~0.25. The coefficient of variation for flow velocity measurements is taken as 0.05~0.10. Take a value of 0.03 to 0.05.

[0204] Considering the material inhomogeneity and measurement errors in physical model experiments, these key parameters (effective cohesion) are... internal friction angle Near-bottom flow velocity Depth of the scour pit at the toe of the slope ), are considered as random variables:

[0205] The mean of the true effective cohesion is a normal distribution. Standard deviation of the normal distribution of true effective cohesion (Coefficient of variation 20%).

[0206] The mean of the internal friction angle is normally distributed. Standard deviation of the normal distribution of internal friction angle (Coefficient of variation 8%).

[0207] The mean of the near-bottom flow velocity follows a normal distribution. Standard deviation of the normal distribution of near-bottom flow velocity .

[0208] It is also considered a random variable, taking into account scanning error, i.e. Instantaneous value of the depth of the scour pit at the toe of the slope The scanning error is randomly distributed within the range.

[0209] This step yields the probability density function and its statistical moments (mean, standard deviation) for each key input parameter, which is the probability distribution model.

[0210] Step S7: Monte Carlo random sampling, based on the probability distribution model defined in step S6, set the total number of simulations. The Mersenne Twister algorithm is used to generate a pseudo-random number sequence. For each simulation... ( ):from Effective cohesion samples are drawn from the distribution for the j-th simulation. ;from Samples of internal friction angles from the j-th simulation are drawn from the distribution. ;from Near-bottom velocity samples are drawn from the distribution for the j-th simulation. ;from Sample the slope toe scour pit depth from the j-th simulation. Composition of the first Group of random parameter vectors . and thus obtain A set of independent random parameter samples.

[0211] Step S8: Batch safety factor calculation, based on the data generated in step S7. Group of random parameter vectors For each group Substituting the values ​​into the formula of step S5, while keeping other geometric parameters unchanged, the safety factor value for the j-th simulation is calculated independently. This process is parallelized, ultimately yielding a sample set of safety coefficients:

[0212]

[0213] A sample set of safety factor distribution containing 10,000 values ​​was obtained.

[0214] This method can systematically analyze the spatial variability and measurement errors of material parameters, determine the probability distribution of key mechanical and hydraulic parameters, and set the number of Monte Carlo simulations for random sampling. By batch calling the improved limit equilibrium equations to calculate multiple sets of safety factor samples, it achieves quantitative analysis of uncertainty propagation while maintaining the physical mechanism, with high computational efficiency and sufficient sample coverage.

[0215] Step S9: Calculate the probability of landslide instability, based on the safety factor sample set obtained in Step S8. The failure domain is defined as The number of samples falling into the failure domain. :

[0216]

[0217] The formula for calculating the probability of landslide instability can also be written as:

[0218] ;

[0219] In the formula: This represents the probability value of landslide instability. This is an indicator function; it takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. The critical stability threshold can be set to 1.0. The total number of simulations for Monte Carlo.

[0220] Get the current time landslide instability probability value .

[0221] Finally, the frequency of landslide instability probability values ​​at the current moment is obtained by statistically analyzing the sample sequence of safety factors that are below the critical stability threshold. This method elevates the deterministic safety factor to a probabilistic risk indicator, which can intuitively reflect the instability probability of landslides under different working conditions, providing a more scientific and reliable quantitative basis for landslide early warning and prevention decisions.

[0222] Step S10: Multi-level early warning output and decision support, analyzing step S5. and step S9 And execute the following judgment logic:

[0223] Level 1 warning signal: If and This indicates that the status is normal.

[0224] Level II warning signal: If or This indicates a level of interest. The system records data and automatically increases the sampling frequency to 20Hz.

[0225] Level 3 Warning Signal: If or This indicates a high alert level. The system issues an audible and visual alarm, warning that "a local shear band has formed, significantly increasing the risk."

[0226] Level IV Warning Signal: If or This indicates a critical level. The system immediately triggers the highest level alarm, determining "overall instability," and based on... Extrapolating the slope of the curve to the remaining settling time (Based on) Curve slope fitting The time corresponding to =1.0 is used to estimate the remaining steady-state time. The real-time display is presented through a visual interface. Probability evolution curves, graded early warning signals, and threshold reports for key disaster-causing factors.

[0227] After issuing a "critical" warning, select the current time. Previous consecutive For each time step, a deterministic safety factor sequence and a landslide instability probability sequence are used to construct nearest-neighbor time windows:

[0228]

[0229] Among them, the current moment For the k-th time step, Indicates the first The deterministic safety factor at each time step, Indicates the first The probability of landslide instability at each time step. Indicates the first The deterministic safety factor at each time step, Indicates the first The probability of landslide instability at each time step.

[0230] Weighted linear fitting was performed on the safety factor sequence and the instability probability sequence respectively:

[0231]

[0232]

[0233] in, The safety factor is the time fitting function. All parameters are fitting parameters for the safety factor; t represents the monitoring time. This is the time-fitting function for the landslide instability probability. All are fitting parameters for the probability of landslide instability;

[0234] when Furthermore, when the goodness of fit meets the preset requirements, the calculated safety factor reaches the critical stability threshold. Predicted time :

[0235]

[0236] when Furthermore, when the goodness of fit meets the preset requirements, the calculated instability probability reaches the critical probability threshold. Predicted time :

[0237]

[0238] Remaining settling time Take the smaller of the two:

[0239]

[0240] when , If the goodness of fit is lower than the preset requirement, the remaining settling time will not be output, and only the message "The trend does not meet the extrapolation conditions" will be output.

[0241] The methods for obtaining the above deterministic safety factor and the warning threshold for the landslide instability probability value are as follows:

[0242] Based on the landslide environment under the dual effects of rainfall infiltration and flood erosion to be assessed, various physical model test systems with different structures and materials are constructed. Tests are continuously conducted under different rainfall amounts until a simulated landslide occurs. Alternatively, tests are set up according to the environmental conditions before an actual landslide event occurs. Multi-source monitoring data during the test process are acquired and preprocessed. The time series of deterministic safety factors and landslide instability probability values ​​for each test are obtained by referring to the method described in this embodiment. Based on the deterministic safety factors and landslide instability probability values ​​at the time of landslide occurrence, alarm thresholds are determined through probabilistic statistical methods. That is, the relationship between deterministic safety factors, landslide instability probability values ​​and alarm levels is determined, thereby obtaining the alarm thresholds for each warning level.

[0243] This invention can achieve the following effects:

[0244] (i) The economic costs are controllable, and it has both significant economic benefits and value for widespread application:

[0245] In-situ multi-sensor monitoring of landslides, which involves both rainfall infiltration and flood erosion, often requires significant investment in manpower, resources, and equipment. Furthermore, the tests are not repeatable and end once a landslide occurs, making it difficult to conduct multiple verifications.

[0246] This approach involves constructing a scaled-down model indoors, using test materials that are identical to or precisely proportioned to simulate the actual landslide, thus keeping material costs relatively controllable. Furthermore, the model can be repeatedly built and reused, ensuring the landslide is in the same initial state before each test, guaranteeing the feasibility of repeated trials and the statistical reliability of the results. Compared to large-scale in-situ data collection, the test conditions are easier to meet, offering advantages such as strong simulation capabilities, ease of use, and low cost.

[0247] (ii) Solving the technical difficulties in conducting field tests and significantly improving the authenticity and reliability of test results:

[0248] For large landslides, conducting full-scale field testing and monitoring typically causes irreversible damage to the surrounding environment, and the testing period is long, with high safety risks, and the damage is difficult to recover after the test. Based on the geological information and relevant physical and mechanical parameters obtained from the landslide site survey, a geometrically similar model with the same scale and materials as the prototype is constructed. Simulation tests are then conducted in a controlled indoor environment, which avoids the environmental damage caused by field tests and can realistically reproduce the instability process of the landslide under the action of external forces such as rainfall, earthquakes, and water level changes.

[0249] From the perspective of similarity principles, using a scaled-down, material-equivalent model ensures that the physical and mechanical properties of the soil and rock mass (such as unit weight, elastic modulus, and shear strength) are strictly consistent with the prototype. This overcomes the distortion of mechanical behavior caused by the size effect in scaled-down models, allowing test results to directly reflect the deformation and failure patterns of landslides at a real scale without the need for conversion and restoration using similarity constants. Compared to existing simple simulation devices, the test conditions provided by this scheme are more consistent with actual engineering conditions, and the test results are more realistic and reliable.

[0250] (III) Enables repeated testing and comparison under multiple operating conditions, significantly improving testing efficiency and accuracy:

[0251] Field prototype tests face numerous uncontrollable factors and often cannot achieve continuous multi-parameter variation tests. The scaled-down, equal-material model testing system proposed in this scheme can simulate various complex landslide-induced scenarios under controlled environmental conditions, primarily in the following ways:

[0252] Single-factor controllable test: It can simulate the impact of single inducing factors such as rainfall infiltration, groundwater seepage, and surcharge preloading on landslide stability, and capture the entire process data of landslide instability in real time through a precise monitoring system.

[0253] Multi-factor coupling test: This system can simultaneously study the coupled effects of factors influencing landslide stability, and realize the study of slope stability by continuous changes in parameters under the dual effects of rainfall infiltration and flood scouring. It eliminates the need for separate tests for different parameters, and even more so for tests of different values ​​of the same parameter, which can greatly save material and time costs and improve test efficiency.

[0254] Through transparent test chambers or high-precision monitoring equipment, the changes in the internal geometry of the soil and rock mass and the deformation of the slope can be clearly observed throughout the entire process from a stable state to instability and sliding, presenting the entire process of landslide evolution in a concise and intuitive way.

[0255] (iv) Obtaining landslide internal response data from multiple dimensions and throughout the entire lifecycle to fill information gaps in on-site observation:

[0256] In actual landslides, key physical quantities such as deep displacement and slip zone stress are difficult to effectively deploy and collect over a long period using sensors. This solution, however, allows for the pre-installation of internal sensors within the landslide body during model construction, enabling the perception of the landslide's internal mechanical behavior.

[0257] By combining devices such as 3D laser scanners and laser displacement sensors, the horizontal displacement and spatial deformation of various components during slope sliding can be measured from multiple dimensions, obtaining deep displacement and plastic zone evolution processes that are difficult to obtain through traditional field monitoring. These data can reveal the complete process of progressive slip zone failure and plastic zone connection, fully reproducing the entire process of landslides from creep to abrupt instability, providing a reliable data foundation for landslide risk assessment and probability determination.

[0258] (v) To provide physical data basis for landslide probability prediction and graded early warning:

[0259] The ultimate goal of this scheme is to assess the actual landslide probability of a landslide body under the combined effects of rainfall infiltration and flood erosion through experimental results using a scaled, isomaterial model. The beneficial effects of this process also include:

[0260] Providing a physically driven basis for probability determination: Traditional landslide probability predictions mostly rely on statistical models and empirical formulas. This scheme, through physical simulation with equal scale and materials, can obtain the physical laws and critical thresholds of landslide instability under various inducing factors such as different rainfall amounts and groundwater level changes, providing real and reliable physical data support for establishing mathematical models of landslide density and characteristic parameters.

[0261] Based on the stress-strain curves of the slip zone and the pore water pressure response law obtained from physical model tests, early warning threshold criteria can be established, and refined graded early warning can be carried out based on different landslide occurrence probabilities.

[0262] Based on the experimental results, the data from the experimental simulation are projected into the actual landslide risk assessment to establish a multi-level early warning threshold for deterministic safety factors and landslide instability probability values.

[0263] In laboratory conditions, this system can be used for emergency monitoring and disaster prevention drills. It also serves as a standardized platform for testing the performance of landslide geological disaster emergency drills and monitoring and early warning systems. Each landslide test iteratively optimizes sensor deployment schemes and early warning mechanisms, improving the efficiency and reliability of applying emerging technologies in landslide disaster prevention. Furthermore, the system provides reliable data calibration and model validation benchmarks for subsequent intelligent monitoring, digital transformation, and landslide model verification in actual landslide projects.

[0264] In addition, in some embodiments, the multi-source monitoring data acquisition system (i.e., integrated monitoring system) of this invention can be directly deployed to the landslide environment site to be assessed, and the same types of data as in step S1 can be collected on site. The data collected on site can be analyzed according to steps S1-S10 to obtain the actual landslide instability probability of the landslide environment to be assessed, and then an early warning can be issued. Similarly, the prediction of the probability of landslide caused by rainfall-flood coupling can be realized, thereby improving the effectiveness and accuracy of landslide risk prevention and control.

[0265] Example 2:

[0266] This embodiment provides a rainfall-flood physical model test system, including:

[0267] Slope Model Box 1 is used to build and contain the soil and rock mass model simulating the slope;

[0268] A rainfall system, erected above the bank slope model box 1, is used to apply artificial rainfall;

[0269] The river flow system is connected to one side of the bank slope model box 1 and is used to simulate the river flow or reservoir water level environment.

[0270] The integrated monitoring system includes multiple sensors distributed inside the bank slope model box 1 and the river flow system, used to collect multi-source monitoring data during the experiment. The multi-source monitoring data includes raw data of the internal physical field, external morphological field and boundary hydraulic field.

[0271] The data processing terminal, connected to the integrated monitoring system, has a built-in processor and memory. The processor is configured to execute the method described in Example 1 to output landslide instability probability prediction results.

[0272] The integrated monitoring system includes:

[0273] The internal physical field monitoring unit includes pore water pressure sensors, earth pressure cells, frequency domain reflectometers, and deep displacement gauges embedded in the slope toe and within the slope model, used to acquire raw data of the internal physical field.

[0274] The external morphology monitoring unit includes a high-speed camera and a three-dimensional laser scanner deployed on the outside of the bank slope model box 1 and the river flow system, which are used to acquire raw data of the external morphology field in real time.

[0275] The boundary hydraulic field monitoring unit includes a flow meter, a water level gauge, and a tipping bucket rain gauge installed in the river flow system to acquire raw boundary hydraulic field data in real time.

[0276] Specifically, such as Figure 1 As shown, this embodiment provides a method for predicting the probability of landslides caused by rainfall-flood coupling based on physical model experiments, including a bank slope model box 1, a rainfall system, a river flow system, and a comprehensive monitoring system. The rainfall system is placed above the bank slope model box 1, and one side of the water tank 12 of the river flow system is connected to the bank slope model box 1. The comprehensive monitoring system is arranged at key parts of each system.

[0277] The bank slope model box 1 is a rectangular box, which is composed of a right side plate, a front side plate, a rear side plate, a bottom plate, and a stainless steel frame. The bottom plate is made of stainless steel, while the right side plate, front side plate, and rear side plate are made of transparent acrylic sheets. The bottom plate and the stainless steel frame are welded together, and the joints of the right side plate, front side plate, and rear side plate are sealed with silicone sealing strips and neutral silicone sealant. A slot is welded at one-third of the distance from the right side plate on the bottom plate for connecting the lifting device. The lifting device consists of a hand-cranked screw jack 6, a lifting rod, and a base 14.

[0278] The rainfall system consists of a water supply module, a rainfall module, and a control module. The water supply module includes a water storage tank 7, a variable frequency water pump 8, and a control valve 10. The rainfall module includes a main water supply pipe 3 (PVC), a branch pipe 4, an atomizing nozzle 5, and a support frame 2 (aluminum alloy). The control module includes an electromagnetic flow meter and a control valve 10.

[0279] The river flow system consists of a water supply module, a water tank 12, and a control module. The water supply module includes a water storage tank 7, a variable frequency water pump 8, water pipes 9, and a control valve 10. The water tank 12 consists of a base plate, side panels, and a stainless steel frame, forming a river model. The base plate is made of stainless steel, and the side panels are made of tempered glass. The base plate and the stainless steel frame are welded together, and the joints of the side panels are sealed with silicone sealing strips and neutral silicone sealant. A slot is welded at one-third of the distance from the water inlet on the base plate for connecting the lifting device. The lifting device consists of a hand-cranked screw jack 6, a lifting rod, and a base 14. Sandpaper is pasted on the walls or bottom of the water tank, and raised particles are arranged to simulate the roughness of a natural river. The control module includes an electromagnetic flow meter and a control valve 10.

[0280] The integrated monitoring system includes a high-speed camera, conventional observation tools (vernier calipers, transparent measuring scales attached to the side panel of the water tank 12 for real-time water level reading), a tipping bucket rain gauge, water flow parameter sensors (acoustic Doppler current meter (ADV), ultrasonic water level gauge), a 3D laser scanner, a rangefinder, a deep displacement meter, a pore water pressure sensor, a frequency domain reflectometer (TDR), an earth pressure cell, and a multi-channel data acquisition instrument.

[0281] The experimental device of this invention has a simple structure and reasonable layout, and can provide a highly similar simulation environment for realizing the evolution of landslide chain disasters induced by rainfall and flood, so as to conduct more effective research on landslides caused by slope toe failure.

[0282] Based on the above technical solution, the present invention can also be optimized and supplemented as follows:

[0283] The front panel of the slope model box 1 is marked with graduations to record the slope deformation in real time. The hand-cranked screw jack 6 is connected to the lifting rod via a coupling. The hand crank is fitted with an anti-slip sleeve, and the rotation ratio is designed to be 1:10 (1cm rise / fall for every 10 rotations) to improve adjustment accuracy. The rainfall module uses one PVC main water pipe 3 with a diameter of 32mm, five branch pipes 4 with a diameter of 15mm (spaced 30cm apart), and four atomizing nozzles 5 (0.5mm orifice diameter, spaced 12cm apart) on each branch pipe 4. The aluminum alloy support frame 2 is height adjustable. The water inlet of the water tank 12 is equipped with a flow straightener 11 (stainless steel wire mesh, aperture 5-10mm) to eliminate water vortices and make the water flow uniform. A baffle 13 is provided at the end of the water tank 12 (to control the water level). One side of the middle of the water tank 12 is connected to the bank slope model box 1. The bottom of the water tank 12 is connected to a lifting device to support bottom slope adjustment (simulating natural river channels with different slopes), with an adjustment accuracy of ±0.1°.

[0284] In the integrated monitoring system, flow meters and water level gauges are installed upstream and downstream of the flume 12 to monitor flow velocity and flow rate; pore water pressure sensors, earth pressure cells, frequency domain reflectometers, and deep displacement gauges are deployed on the slope and toe of the soil and rock mass model to monitor the sliding state of the bank slope; tipping bucket rain gauges are installed below the rainfall module; high-speed cameras are arranged on the outside of the observation surfaces of the flume 12 and the bank slope model box 1, and a three-dimensional laser scanner is deployed opposite the toe of the slope to observe the morphological changes of the toe of the slope in real time; finally, the data are integrated into a multi-channel data acquisition instrument.

[0285] Combined with appendix Figure 1 As shown, the test steps of the test system in this embodiment are as follows:

[0286] Step 1: Connect the bank slope model box 1 to the hand-cranked screw jack 6, place the lifting rod in the slot at the bottom of the bank slope model box 1, and adjust the hand-cranked screw jack 6 to place the bank slope model box 1 at the required tilt angle for the test to simulate the natural bank slope; connect the water tank 12 to the hand-cranked screw jack 6, place the lifting rod in the slot at the bottom of the bank slope model box 1, and adjust the hand-cranked screw jack 6 to place the water tank 12 at the required tilt angle for the test to simulate the natural river channel inclination.

[0287] Step 2: Connect the adjusted bank slope model box 1 to the water tank 12, and seal the joint with silicone sealing strip and neutral silicone sealant.

[0288] Step 3: Arrange the soil and rock mass in layers in the bank slope model box 1. Select natural soil and artificially prepared similar materials to simulate different soil and rock layers, including the lower bedrock layer, the middle permeable layer, and the upper surface layer. The bedrock layer material is prepared by mixing quartz sand, cement, kaolin, and water according to the mass ratio. The middle permeable layer material is prepared by mixing engineering sand, gypsum powder, and water according to the mass ratio. The upper surface layer material is simulated by using natural soil that meets the particle size distribution conditions, and the dry density and moisture content of each simulated soil and rock mass are strictly controlled.

[0289] Step 4: Arrange a rainmaking device above the bank slope model box 1. The rainmaking device is constructed with an aluminum alloy frame. The main water supply pipe 3 and branch pipe 4 are installed on the aluminum alloy frame and connected with T-shaped and L-shaped interfaces. Four atomizing nozzles 5 are arranged at equal intervals on each branch pipe 4. The water source for simulating rainfall is provided by the water storage tank 7. The water channel is from the water storage tank 7 through the frequency conversion water pump 8, water pipe 9, control valve 10, water pipe 9, main water supply pipe 3, and branch pipe 4 to the atomizing nozzles 5.

[0290] Step 5: Arrange a partition 13 at the end of the water tank 12. The partition 13 can be placed in the slot of the stainless steel frame. Weld a stainless steel wire mesh as a flow rectifier 11 at the inlet of the water tank 12. The river water source is provided by the water storage tank 7. The water channel flows from the water storage tank 7 through the frequency conversion water pump 8, water pipe 9, control valve 10, and water pipe 9 to the water tank 12.

[0291] Step 6: Deploy monitoring equipment; bury pore water pressure sensors and soil pressure cells inside the slope toe, bury frequency domain reflectometers and deep displacement gauges in the slope, deploy a rangefinder outside the slope toe and connect it to a multi-channel data acquisition system; deploy a flow meter and water level gauge at the bottom of the water tank 12 and 20cm upstream of the slope toe and connect them to the multi-channel data acquisition system; deploy high-speed cameras on one side of the bank slope model box 1 and on the outside of the water tank 12, and deploy a 3D laser scanner and a rangefinder on the opposite side of the slope toe, i.e., outside the water tank 12 and connect them to the multi-channel data acquisition system.

[0292] Step 7: Debug all systems to ensure the normal operation of the experiment. Data is collected throughout the process using the integrated monitoring system, and the experimental data is processed through the data processing terminal to output the current landslide instability probability value and the corresponding warning level. This method elevates the deterministic safety factor to a probabilistic risk indicator, which can intuitively reflect the instability probability of landslides under different working conditions, providing a more scientific and reliable quantitative basis for landslide early warning and prevention decisions.

[0293] This setup enables more accurate measurement of slope deformation, improves the adjustment precision of the elevator, provides more uniform and reasonable simulation of rainfall, more realistic simulation of river flow, and multi-directional monitoring to enhance the completeness of data monitoring.

[0294] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0295] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0296] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature being directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature being directly above or diagonally above the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0297] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not presuppose that they refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0298] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A method for predicting the probability of landslides caused by rainfall-flood coupling based on physical model experiments, characterized in that, Includes the following steps; Step S1: Construct a physical model test system to simulate the landslide environment to be evaluated, and conduct simulation tests based on the physical model test system. Acquire and preprocess multi-source monitoring data during the test process. The multi-source monitoring data includes raw data of internal physical field, external morphological field and boundary hydraulic field. Construct a standardized multi-dimensional time series input matrix. Step S2: Calculate the effective normal stress sequence at the sliding section in real time based on the input matrix; Step S3: Construct a nonlinear reduction model for the strength of the soil and rock mass, and dynamically invert the time-varying strength parameters that change with time; Step S4: Update the geometric topology of the landslide body in real time to obtain the updated strip geometric parameters; Step S5: Based on the effective normal stress sequence, time-varying intensity parameters, and updated strip geometric parameters at the sliding zone, the deterministic safety factor at any time is calculated according to the improved limit equilibrium state equation considering the action of hydrodynamic pressure and changes in geometric shape. Step S6: Analyze the spatial variability and measurement error of material parameters, and determine the probability distribution of key mechanical and hydraulic parameters; Step S7: Set the number of Monte Carlo simulations, and randomly sample based on the probability distribution of the determined key mechanical and hydraulic parameters to generate multiple sets of random parameter sample sets; Step S8: Substitute each set of random parameter samples into the improved limit equilibrium state equation to calculate the corresponding safety factor sample sequence in batches; Step S9: Count the frequency of the safety factor sample sequence that is less than the critical stability threshold, and calculate the landslide instability probability value at the current moment.

2. The rainfall-flood coupled landslide probability prediction method according to claim 1, characterized in that, In step S1, the construction process of the standardized multidimensional time-series input matrix includes: Acquire internal physics data sequence ,in, for The pore water pressure measured by the time sensor. for Total earth pressure measured by the sensor at any given time. for Soil saturation measured by a time sensor; Acquire external morphological field data sequence ,in, for The depth of the scour pit at the foot of the slope at any given moment; Obtain boundary hydraulic field data sequence ,in, for The flow rate at any given moment. for The water level at any given time; A sliding window filtering algorithm is used to remove high-frequency noise, and linear interpolation alignment is performed based on millisecond-level timestamps to form a standardized multidimensional time-series input matrix. : ； In this context, the superscript T indicates matrix transpose.

3. The rainfall-flood coupled landslide probability prediction method according to claim 2, characterized in that, In step S2, the formula for calculating the effective normal stress sequence is: ； In the formula: express The effective normal stress at the constant sliding section; express Total earth pressure measured by the time sensor; express Pore ​​water pressure measured by a time sensor; This represents the effective stress coefficient, which is taken as 1.0 for saturated soil and according to... Linear interpolation is determined; In step S3, the time-varying intensity parameters include effective cohesion and effective internal friction angle; The expression for the nonlinear reduction model of soil and rock strength is as follows: ； ； In the formula: express Effective cohesion at any given time; express Effective internal friction angle at any given time; These represent the initial effective cohesion and initial effective internal friction angle of the model material, respectively; express The soil saturation at time t, with a value range of [0,1]. express The depth of scour at the toe of the slope at any given moment; This indicates the initial height of the bank slope model; Represents the natural exponential function; All of these represent the material damage coefficient, calibrated by pre-testing; This represents a nonlinear exponent.

4. The rainfall-flood coupled landslide probability prediction method according to claim 1, characterized in that, In step S4, the geometric topology of the landslide body is updated in real time to obtain the updated strip geometric parameters, including: Reconstructing 3D point cloud data using a 3D laser scanner Slope surface equation at time Identify the scour boundary at the toe of the slope; The landslide mass is divided along the potential sliding surface into... The nth vertical strip, for the nth Individual blocks: Update # The bottom arc length of each strip : In the initial state, the first The bottom arc segment of each strip Based on the time... Reconstructing the scour airspace at the foot of the slope from 3D point cloud data and the initial slope area After deducting the scour void, the current remaining soil area is obtained. : ； Intersect the initial bottom slip arc segment of the i-th block with the current remaining soil region to obtain the effective slip arc segment that is still in contact with the soil. : ； The updated bottom arc length of the i-th block is: ； When the scour airspace overlaps with the bottom sliding arc segment of the i-th block, the sliding arc segment corresponding to the overlapping part is determined to be an exposed segment or a suspended segment, and is no longer included in the effective anti-slip arc length. In the formula, This refers to the slope body in its initial state. For a moment The slope toe scour airspace identified based on 3D point cloud data; For a moment The current remaining soil area after deducting the scoured airspace; For the initial state, the first The bottom arc segment of each strip; For a moment No. The effective bottom slip arc segment that is still in contact with the remaining soil area; To find the intersection; For a moment No. The updated effective bottom arc length of each strip block; Operator for calculating curve length; For monitoring time; Number the blocks. , The total number of blocks; No. Volume of each strip The update method is as follows: ； in, For the first Each block at time The horizontal projection area The elevation of the current slope surface is obtained from the reconstruction of the 3D point cloud. For potential slip surface elevation, This means that if the value inside the parentheses is greater than 0, the original value is used; otherwise, 0 is used. Update # Each block's weight The calculation formula is: ； In the formula: express Saturation intensity at any given moment; Indicates natural severe; Indicates the first The volume of each strip; Indicates the first Soil saturation of each block.

5. The rainfall-flood coupled landslide probability prediction method according to claim 4, characterized in that, In step S5, based on the effective normal stress sequence, time-varying intensity parameters, and updated strip geometric parameters at the slip zone, and according to the constructed improved limit equilibrium state equation considering hydrodynamic pressure and geometric changes, the deterministic safety factor at any given time is calculated, including: Calculate the first Anti-slip moment of each strip for: ； in, Let be the radius of the sliding arc. For the first Effective sliding arc length at the bottom of each strip block For the first Calculate the width of each strip; No. Effective force of each block normal direction The following was calculated based on the effective normal stress sequence at the slip zone: ； In the formula, The effective normal stress sequence at the slip zone The first obtained through spatial mapping or interpolation The average effective normal stress at the bottom of each strip, when the effective normal stress of the sliding strip is... When multiple monitoring points provide data together, the average effective normal stress is expressed as: In the formula, For the first Each slip zone monitoring point at time The effective normal stress; For the first The monitoring point for the first Spatial interpolation weights for each block, and satisfying ; Calculate the first The downward torque of each block : ； Among them, the first The dynamic water pressure of each block The calculation is as follows: ； Calculate the deterministic safety factor: ； In the formula: : No. Inclination angle of the strip surface; : Coefficient of drag of an object : Water density, n is the total number of strips; : No. The projected area of ​​the strip in the direction of water flow. : The angle between the water flow vector and the tangent of the sliding surface.

6. The rainfall-flood coupled landslide probability prediction method according to claim 5, characterized in that, In step S6, the spatial variability of material parameters and measurement errors are analyzed to determine the probability distribution of key mechanical and hydraulic parameters, including: Effective cohesion internal friction angle Near-bottom flow velocity The probability distribution type is determined to be either normal or log-normal, and its statistical characteristic values ​​are determined as follows: Mean of normal or log-normal distribution Instantaneous values ​​of effective cohesion, internal friction angle, and near-bottom velocity derived from inversion; standard deviation Based on the coefficient of variation of the material homogeneity test Sure: ; The coefficient of variation of effective cohesion The coefficient of variation of the internal friction angle is taken as 0.15~0.

25. The coefficient of variation for near-bottom flow velocity is taken as 0.05~0.

10. Take a value of 0.03~0.05; Define the depth of the scour pit at the toe of the slope. The depth of the scour pit at the toe of the slope at this moment is randomly distributed within the scanning error range.

7. The rainfall-flood coupled landslide probability prediction method according to claim 6, characterized in that, In step S7, the number of Monte Carlo simulations is set, and random sampling is performed based on the probability distributions of the determined key mechanical and hydraulic parameters to generate multiple sets of random parameter samples, including: Set the total number of simulations ; Use the Mersenne Twister algorithm to generate a pseudo-random number sequence; For the j-th simulation: Generate a random number seed and draw the first random number from the defined distribution. Group of random parameter vectors ; in, This is the effective cohesion sample for the j-th simulation; This is the sample of the internal friction angle in the j-th simulation; This is the near-bottom velocity sample from the j-th simulation; This is a sample of the slope toe scour pit depth from the j-th simulation. In step S8, each set of random parameter samples is substituted into the improved limit equilibrium state equation to calculate the corresponding safety factor sample sequence in batches, including: The first Group of random parameter vectors Substituting into the formula for calculating the deterministic safety factor, we obtain the first... Safety factor of this simulation Finally, a safety factor sample set is obtained. .

8. The rainfall-flood coupled landslide probability prediction method according to claim 7, characterized in that, In step S9, the frequency of values ​​below the critical stability threshold in the statistical safety factor sample sequence is counted, and the landslide instability probability value at the current moment is calculated, including: landslide instability probability value The calculation formula is: ； In the formula: This is an indicator function; it takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. This is the critical stability threshold; The total number of simulations for Monte Carlo.

9. The rainfall-flood coupled landslide probability prediction method according to claim 8, characterized in that, The method further includes: Based on the landslide instability probability value and the deterministic safety factor, a preset graded early warning rule is matched, and multi-level early warning signals are output. The tiered early warning rules include: like and It outputs a first-level warning signal to indicate a safe state; like or It outputs a level-two early warning signal to alert users and increases the monitoring frequency. like or It outputs a three-level early warning signal to indicate the risk of local shear damage; like or It outputs a level four early warning signal to alert that a landslide is about to become unstable, and simultaneously outputs the remaining stabilization time extrapolated from the current trend.

10. A rainfall-flood physical model test system, characterized in that, include: A bank slope model box is used to build and contain a soil and rock mass model simulating a bank slope. A rainfall system, erected above the bank slope model box, is used to apply artificial rainfall; The river flow system is connected to one side of the bank slope model box and is used to simulate the river flow or reservoir water level environment. The integrated monitoring system includes multiple sensors distributed inside the bank slope model box and in the river flow system, used to collect multi-source monitoring data during the experiment. The multi-source monitoring data includes raw data of the internal physical field, external morphological field and boundary hydraulic field. A data processing terminal, connected to the integrated monitoring system, has a built-in processor and memory, the processor being configured to execute the method as described in any one of claims 1-9 to output a landslide instability probability prediction result; The integrated monitoring system includes: The internal physical field monitoring unit includes pore water pressure sensors, earth pressure cells, frequency domain reflectometers, and deep displacement gauges embedded in the slope toe and within the slope model, used to acquire raw data of the internal physical field. The external morphology monitoring unit includes a high-speed camera and a 3D laser scanner deployed on the outside of the bank slope model box and the river flow system, which are used to acquire raw data of the external morphology field in real time. The boundary hydraulic field monitoring unit includes a flow meter, a water level gauge, and a tipping bucket rain gauge installed in the river flow system to acquire raw boundary hydraulic field data in real time.

Images