Landslide stability assessment method and system combining material point method and transfer coefficient method

Abstract

A landslide stability assessment method and system combining the material point method and the transfer coefficient method includes: constructing a three-dimensional material point model of the landslide body based on the material point method to obtain the stress field distribution, deformation characteristics, and fluid-solid coupling effects of the landslide body; dividing the landslide body into multiple blocks and determining the block area and location of each material point; constructing a stress transfer matrix based on the location and physical information of the material points; calculating the normal stress perpendicular to the contact surface between each block area of the landslide body and the tangential stress parallel to the contact surface between each block area based on the stress transfer matrix, and calculating the real-time stability coefficient; and calculating the dynamic safety factor of the landslide body based on displacement, pore water pressure and stress, and the real-time stability coefficient, and performing a stability assessment on the landslide body. By combining the material point method and the transfer coefficient method, the present invention considers the mechanical behavior of the landslide body and the material propagation process, thereby improving the accuracy of landslide stability assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of landslide prevention and control engineering, and in particular to a landslide stability assessment method and system combining a material point method with a transfer coefficient method. Background Art

[0002] Currently, landslide stability assessment relies on traditional mechanical models, such as the limit equilibrium method and the finite element method. However, these methods have limitations, particularly in complex geological conditions, and cannot accurately predict landslide occurrence. While the material point method and transfer coefficient method have their advantages, they struggle to achieve high-precision, real-time early warnings due to the difficulty of parameter optimization and high computational complexity.

[0003] A Chinese patent application, publication number CN108548730A, published on September 18, 2018, discloses a landslide stability assessment method based on the transfer coefficient method and surface displacement. The method includes dividing the landslide to be assessed into multiple vertical sliding blocks, determining the stress state of each sliding block, calculating the shear displacement of the sliding band at the bottom of each sliding block, and calculating the landslide stability coefficient based on the sliding band shear displacement. Although this method, based on the transfer coefficient method and measured landslide surface displacement data, establishes a relationship between landslide surface displacement and landslide stability, and can use displacement data to assess landslide stability, thus better facilitating real-time landslide monitoring, it still suffers from the following drawbacks: This method only considers the mechanical properties of the landslide body and does not consider the material propagation process within the landslide body, resulting in low landslide stability assessment accuracy. Therefore, a new landslide stability assessment method is needed that comprehensively considers both the mechanical properties of the landslide body and the material propagation process to improve the accuracy of landslide stability assessment. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art and to provide a landslide stability assessment method and system combining the material point method with the transfer coefficient method to improve the assessment accuracy of landslide stability.

[0005] To achieve the above objectives, the technical solution of the present invention is:

[0006] In a first aspect, the present invention provides a landslide stability assessment method combining a material point method and a transfer coefficient method, comprising:

[0007] Based on the material point method, a three-dimensional material point model of the landslide body is constructed to obtain the stress field distribution, deformation characteristics and fluid-solid coupling effect of the landslide body;

[0008] Divide the landslide body into multiple block areas and determine the block areas and positions of each material point;

[0009] A stress transfer matrix is constructed with the location and physical information of material points as elements. Based on the stress transfer matrix, the normal stress in the vertical direction between each block area of the landslide body and the tangential stress parallel to the contact surface between each block area are calculated. The real-time stability coefficient is calculated using the contact force calculated from the normal stress and tangential stress.

[0010] The dynamic safety factor of the landslide is calculated based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability coefficient obtained from the transfer coefficient method, and the stability of the landslide is evaluated based on the dynamic safety factor.

[0011] Preferably, the dynamic safety factor is:

[0012] ;

[0013] Where, express Dynamic safety factor at the moment; express Real-time stability coefficient at each moment; Indicates the influence coefficient of the real-time stability coefficient; express displacement of moments; Indicates a time interval; Indicates the influence coefficient of deformation rate; express Pore water pressure at the moment; Indicates the influence coefficient of pore water pressure; express The stress gradient norm at time t; Represents the influence coefficient of stress gradient.

[0014] Preferably, the construction of a stress transfer matrix with the positions and physical information of material points as elements includes:

[0015] ;

[0016] ;

[0017] ; ;

[0018] ; ;

[0019] ;

[0020] ; ;

[0021] ; ;

[0022] ;

[0023] Where, Represents the stress information matrix with the position and physical information of material points as elements; Indicates the position of a material point; Indicates the The coordinates of the points; Represents the physical information of a material point; Indicates the Physical information of each point; Represents the stress information matrix after transmission; Indicates the position of the material point after transfer; Represents the physical information of the material point after transfer; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; 、 、 Respectively represent The first block area The density, volume, and stress information carried by each material point.

[0024] Preferably, the calculation formulas for the normal stress in the vertical direction between the block areas of the landslide body and the tangential stress in the direction parallel to the contact surface between the block areas are as follows:

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] Where, represents the stress tensor; 、 、 Respectively 、 、 principal stresses in the direction; 、 、 Respectively 、 、 Tangential stress in the direction; represents the boundary normal vector; 、 、 Respectively represent the components of the boundary normal vector in each direction; represents the normal component; represents the tangential component; represents the transpose of the boundary normal vector; It represents the normal stress in the vertical direction between the blocks of the landslide body; It represents the tangential stress parallel to the contact surface between the various blocks of the landslide body; It represents the contact area of each block of the landslide body.

[0032] Preferably, historical landslide data is reconstructed based on a generative adversarial network model, specifically including:

[0033] A physics-aware generative adversarial network model was constructed, with a geomechanical equation solver embedded in the generator to generate landslide data that conforms to physical constraints. A dual-channel discriminator was designed, with the first channel assessing the authenticity of the generated landslide data and the second channel verifying the physical consistency of the stress field divergence and pore water pressure gradient.

[0034] A hierarchical physical loss function is constructed, which includes global momentum conservation constraints and local contact surface stress distribution constraints. The weights of the hierarchical physical loss function are dynamically adjusted according to the real-time monitoring data of the landslide body, resulting in an optimized physical-aware generative adversarial network model.

[0035] Historical landslide data are reconstructed based on the optimized physical-aware generative adversarial network model.

[0036] Preferably, the layered physical loss function is:

[0037] ;

[0038] ;

[0039] Where, represents the global physical loss function; represents the weight of the error term in the equilibrium equation; represents the divergence of the stress tensor; represents the stress tensor used to generate the landslide data; Indicates density; represents the acceleration due to gravity; represents the weight of the error term of the shear failure criterion; represents the shear stress tensor used to generate the landslide data; Indicates cohesion; represents the normal component; represents the internal friction angle; represents the local boundary physical loss function; represents the weight of the local stress rationality item; represents the boundary normal vector; Represents the weight of the stress tensor symmetry and projection error term;

[0040] The method of dynamically adjusting the weight of the layered physical loss function according to the real-time monitoring data of the landslide body includes:

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] Where, represents the momentum weight decay rate; represents the deformation rate; represents the pore water pressure gradient; represents the rate of change of the boundary normal vector; It represents the sensitivity coefficient of stress symmetry of complex terrain; represents the gradient of the sliding surface direction field; 、 、 、 are the initial values of each weight.

[0047] Preferably, for loose soil landslides, the area with a stress gradient greater than or equal to 15 kPa or a displacement deformation rate greater than or equal to 2 mm / h is divided into a local complex refinement area;

[0048] For rock landslides, the area with a stress gradient greater than or equal to 20 kPa or a displacement deformation rate greater than or equal to 1.5 mm / h is divided into a local complex refinement zone;

[0049] For local complex and refined areas, a three-dimensional material point model of the landslide body is constructed based on the material point method.

[0050] In a second aspect, the present invention provides a landslide stability assessment system combining a material point method and a transfer coefficient method, the system being applied to the above-mentioned method, the system comprising:

[0051] A 3D material point model construction module is used to construct a 3D material point model of the landslide based on the material point method to obtain the stress field distribution, deformation characteristics, and fluid-solid coupling effects of the landslide;

[0052] The real-time stability coefficient calculation module is used to divide the landslide into multiple blocks and determine the block area and position of each material point. It constructs a stress transfer matrix with the position and physical information of the material points as elements, and calculates the normal stress in the vertical direction between the blocks of the landslide and the tangential stress parallel to the contact surface between the blocks based on the stress transfer matrix. The real-time stability coefficient is calculated based on the contact force calculated from the normal stress and tangential stress.

[0053] The stability assessment module is used to calculate the dynamic safety factor of the landslide based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability coefficient obtained from the transfer coefficient method, and to perform stability assessment on the landslide based on the dynamic safety factor.

[0054] In a third aspect, the present invention provides a landslide stability assessment device combining a material point method and a transfer coefficient method, comprising a memory and a processor;

[0055] The memory is configured to store computer program code and transmit the computer program code to the processor;

[0056] The processor is configured to execute the method described above according to instructions in the computer program code.

[0057] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the method described above when executed by a processor.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention provides a landslide stability assessment method and system combining the material point method with the transfer coefficient method. By combining the material point method and the transfer coefficient method, the mechanical behavior of the landslide body and the material propagation process are comprehensively considered, thereby improving the accuracy of landslide stability assessment. Furthermore, by designing a stress transfer matrix, local stress data obtained by the material point method is converted into input parameters for the transfer coefficient method. This ensures information synchronization between the material point method and the transfer coefficient method at both the local and global levels, ensuring more accurate stability assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 The present invention provides a flow chart of a landslide stability assessment method combining a material point method with a transfer coefficient method.

[0061] Figure 2 This is a structural block diagram of the physical perception generative adversarial network model provided by an embodiment of the present invention.

[0062] Figure 3 It is a structural block diagram of a landslide stability assessment system combining a material point method and a transfer coefficient method provided by the present invention.

[0063] Figure 4 It is a structural block diagram of the landslide stability assessment device provided by the present invention that combines the material point method with the transfer coefficient method. DETAILED DESCRIPTION

[0064] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] See also Figure 1 The present invention provides a landslide stability assessment method combining a material point method and a transfer coefficient method, comprising:

[0066] S1. Based on the material point method, a three-dimensional material point model is constructed for the landslide body to obtain the stress field distribution, deformation characteristics and fluid-solid coupling effect of the landslide body.

[0067] The method constructs a three-dimensional material point model of the landslide body based on the material point method to obtain the stress field distribution, deformation characteristics and fluid-solid coupling effect of the landslide body, including: discretizing the landslide body into multiple material points with mass, velocity and stress state; using the Lagrangian framework to simulate the interaction between the material points; and calculating the dynamic behavior of each material point to analyze the stress field distribution, deformation characteristics and fluid-solid coupling effect of the landslide body.

[0068] This method discretizes the landslide into multiple material point units, each containing mass, position, velocity, and stress tensor, to simulate the complex dynamic response of the landslide. A material point method mathematical model is constructed, utilizing mesh-particle coupling to calculate the stress field distribution and deformation evolution of the landslide. Combined with fluid-structure coupling theory, the method simulates the influence of internal fluids (such as pore water pressure) on landslide stability. This method can provide detailed local stress and deformation data.

[0069] S2. Divide the landslide body into multiple block areas and determine the block areas and positions of each material point; construct a stress transfer matrix with the position and physical information of the material points as elements, and calculate the normal stress in the vertical direction between each block area of the landslide body and the tangential stress parallel to the contact surface between each block area based on the stress transfer matrix, and calculate the real-time stability coefficient through the contact force calculated by the normal stress and tangential stress.

[0070] This method uses the transfer coefficient method for real-time stability assessment. It divides the landslide into multiple sub-regions, calculates the normal and tangential stresses at the bottom contact surfaces of each sub-region, analyzes the mechanical coupling between the different sub-regions, and analyzes the material migration, energy transfer, and geological evolution processes within the landslide, thereby analyzing the overall stability of the landslide.

[0071] In the specific implementation, the partition and position of each point are screened according to the set block area. The specific identification method is: first, set the 1st to The coordinate threshold of the partition; secondly, by 、 、 The specific partition positions are identified in the order of . Since reasonable block division can not only improve the calculation efficiency but also ensure the accuracy and reliability of the results, the general division basis can be composed of the following aspects: the geometric shape and geological characteristics of the landslide body, the potential sliding surface, the boundary conditions and external influences, and the calculation accuracy requirements.

[0072] During the coupling process between the material point model and the transfer coefficient method, a stress transfer matrix is designed to convert the local stress data obtained from the material point method into input parameters for the transfer coefficient method. This ensures that the two methods synchronize information at the local and global levels, ensuring more accurate stability assessment.

[0073] S3. Calculate the dynamic safety factor of the landslide based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability coefficient obtained from the transfer coefficient method, and evaluate the stability of the landslide based on the dynamic safety factor.

[0074] Furthermore, the dynamic safety factor is:

[0075] ;

[0076] Where, express Dynamic safety factor at the moment; express Real-time stability coefficient at each moment; Indicates the influence coefficient of the real-time stability coefficient; express displacement of moments; Indicates a time interval; Indicates the influence coefficient of deformation rate; express Pore water pressure at the moment; Indicates the influence coefficient of pore water pressure; express The stress gradient norm at time t; Represents the influence coefficient of stress gradient.

[0077] Calculate the real-time stability coefficient in the three-dimensional transfer coefficient method section The physical information such as the landslide friction coefficient used in the landslide is obtained through geological exploration data, and the inclination of each block is The potential sliding surface inferred from the large deformation of the material point method can be further calculated in combination with the drilling data. The above-mentioned influence coefficients can be weighted by using methods such as grey correlation or Pearson correlation coefficient method and applying landslide characteristics.

[0078] Furthermore, the construction of the stress transfer matrix with the position and physical information of the material points as elements includes:

[0079] ;

[0080] ;

[0081] ; ; ;

[0082] ; ; ;

[0083] ;

[0084] ; ;

[0085] ; ;

[0086] ;

[0087] Where, Represents the stress information matrix with the position and physical information of material points as elements; Indicates the location of a material point; Indicates the The coordinates of the points; Represents the physical information of a material point; Indicates the Physical information of each point; Represents the stress information matrix after transmission; Indicates the position of the material point after transfer; Represents the physical information of the material point after transfer; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; 、 、 Respectively represent The first block area The density, volume, and stress information carried by each material point.

[0088] Furthermore, the calculation formulas for the normal stress in the vertical direction between the block areas of the landslide body and the tangential stress in the direction parallel to the contact surface between the block areas are as follows:

[0089] ;

[0090] ;

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] Where, represents the stress tensor; 、 、 Respectively 、 、 principal stresses in the direction; 、 、 Respectively 、 、 Tangential stress in the direction; represents the boundary normal vector; 、 、 Respectively represent the components of the boundary normal vector in each direction; represents the normal component; represents the tangential component; represents the transpose of the boundary normal vector; It represents the normal stress in the vertical direction between the blocks of the landslide body; It represents the tangential stress parallel to the contact surface between the various blocks of the landslide body; It represents the contact area of each block of the landslide body.

[0096] For further information, see Figure 2 , reconstructing historical landslide data based on the generative adversarial network model, specifically including:

[0097] A physics-aware generative adversarial network model was constructed, embedding a module for solving geomechanical equations (Mohr-Coulomb shear strength criterion and Biot fluid-structure coupling equation) in the generator to generate high-resolution landslide data that complies with physical constraints (stress field conservation and shear strength criterion). A dual-channel discriminator was designed: the first channel evaluates the authenticity of the generated landslide data, and the second channel verifies the physical consistency of the stress field divergence and pore water pressure gradient.

[0098] A hierarchical physical loss function is constructed, which includes global momentum conservation constraints and local contact surface stress distribution constraints. The weights of the hierarchical physical loss function are dynamically adjusted according to the real-time monitoring data of the landslide body, resulting in an optimized physical-aware generative adversarial network model.

[0099] Historical landslide data are reconstructed based on the optimized physical-aware generative adversarial network model.

[0100] The present invention constructs a physics-aware generative adversarial network (GAN) model by embedding physical constraints within the original adversarial network model framework. This primarily involves improvements to the generator and discriminator. Specifically, the generator inputs information related to displacement, stress, and pore water pressure from historical landslide data, along with information on different geological characteristics such as slope and lithology. A physical-aware module within the geomechanical equation solver, added to the decoding layer, compares the real-time generated stress field divergence and pore water pressure gradient with the actual physical field. Second, physical field verification and identification are performed using real or generated data. The first channel evaluates the authenticity of the generated data, while the second channel verifies the physical consistency of the stress field divergence and pore water pressure gradient. GAN can generate high-quality landslide data that complies with physical constraints based on geomechanical equations, addressing the shortcomings of insufficient or poor-quality actual monitoring data. It not only generates standardized datasets but also high-resolution data, thereby improving the accuracy of landslide models. The generator embeds geomechanical equations to ensure that the generated data complies with physical constraints, such as stress field conservation and shear strength criteria, thus avoiding the problem of data generation inconsistent with physical laws in traditional methods.

[0101] Furthermore, the layered physical loss function is:

[0102] ;

[0103] ;

[0104] Where, represents the global physical loss function, which is used to control the macroscopic static balance and sliding strength rationality; represents the weight of the error term in the equilibrium equation; represents the divergence of the stress tensor; represents the stress tensor used to generate the landslide data; Indicates density; represents the acceleration due to gravity; represents the weight of the error term of the shear failure criterion; represents the shear stress tensor used to generate the landslide data; Indicates cohesion; represents the normal component; represents the internal friction angle; Represents the local boundary physical loss function, which is used to control the rationality of the local surface stress projection and the physical properties of the tensor structure; represents the weight of the local stress rationality item; represents the boundary normal vector; The physical loss function is a metric that quantifies the difference between the model's predicted physical quantities (such as stress field, stress gradient, and Mohr-Coulomb failure criterion) and the ideal physical constraints or observed data.

[0105] Among them, the error term of the equilibrium equation , indicating that stress divergence should be balanced with gravity load, reflecting whether the dynamic equilibrium equation is satisfied; it is used to measure the degree of deviation from "overall static equilibrium" in the numerical results. The more rapidly a landslide develops, the more the model focuses on localized slip failures, reducing rigid constraints on global equilibrium to enhance model adaptability.

[0106] Shear failure criterion error term , represents the deviation of the Mohr-Coulomb failure criterion, reflecting whether the material meets the failure conditions; Indicates the actual failure strength envelope; used to strengthen the rationality of failure criteria at key surfaces such as sliding surfaces and weak surfaces. The greater the pore water pressure gradient, The larger the pore water pressure gradient, the smaller the The smaller.

[0107] Local stress rationality item , which means that after the stress tensor is projected onto the sliding surface, it should match the actual sliding surface stress; it is used to measure the generated stress vector and the interface normal component The difference between the two can improve the accuracy of stress simulation on key failure surfaces. The more complex the local structural surface changes (such as rock folds, fractures, loose boundaries), The larger it is, the stronger the constraint effect of the normal stress control term.

[0108] Stress tensor symmetry and projection error terms , used to simulate the stress tensor should be close to the symmetric projection of the normal stress of the sliding surface, check whether the generated stress tensor shows a tensor structure dominated by the normal stress near the sliding surface; used to limit non-physical shear stress distortion or excessive tensor distortion. The more asymmetric the stress, the more it indicates the presence of numerical noise or non-physical stress structure, and the need to increase this penalty, The robustness of the model is enhanced, especially in discontinuous areas such as interfaces and damage zones, showing stronger stable adjustment capabilities.

[0109] The method of dynamically adjusting the weight of the layered physical loss function according to the real-time monitoring data of the landslide body includes:

[0110] ;

[0111] ;

[0112] ;

[0113] ;

[0114] ;

[0115] Where, Indicates the momentum weight decay rate, ranging from 5 to 10; Indicates the deformation rate and reflects the degree of dynamics; represents the pore water pressure gradient; represents the rate of change of the boundary normal vector; It represents the stress symmetry sensitivity coefficient of complex terrain, with a value of 20; Represents the gradient of the sliding surface direction field (the more complex, the greater the change); 、 、 、 are the initial values of each weight, for a stable slope, The value is 50. The value is 1. The value is 3. Take the value as 0.1; for significant deformation, The value range is 20 to 30. The value is 3. The value is 5. The value ranges from 0.05 to 0.1.

[0116] By combining GAN with real-time monitoring data to dynamically adjust weights and loss functions, the generated data can promptly reflect the latest dynamics of the landslide, greatly enhancing the reliability and real-time performance of landslide stability analysis. Specifically, momentum conservation is prioritized in areas with high displacement rates, while shear strength constraints are strengthened in areas with high pore water pressure gradients.

[0117] The present invention optimizes model parameters through machine learning algorithms, automatically learning the potential laws of landslide occurrence from historical data, and at the same time using generative adversarial network (GAN) technology to effectively supplement problems such as missing monitoring data and insufficient boundary conditions in landslide stability analysis, making the model more complete and reliable, avoiding analysis result errors caused by insufficient data or quality problems, and improving the overall adaptability and accuracy of the model.

[0118] Furthermore, the present invention also provides a multi-scale collaborative analysis strategy, which processes the local complex stress field and overall stress transfer effect of the landslide body by combining a local refined grid with a large-scale coarse-scale grid; couples the high-precision data generated by GAN at different scales to supplement boundary conditions or missing monitoring data; and integrates the dynamic feedback of the multi-scale landslide model to improve the model's ability to predict the stability of the landslide body over a long period of time.

[0119] The multi-scale collaborative analysis strategy adopts dynamic adaptive grid division. For loose soil landslides, the area with stress gradient greater than or equal to 15kPa or displacement deformation rate greater than or equal to 2mm / h is divided into local complex and refined areas; for rock landslides, the area with stress gradient greater than or equal to 20kPa or displacement deformation rate greater than or equal to 1.5mm / h is divided into local complex and refined areas, otherwise it is set as a coarse-scale observation area.

[0120] Specifically, the coarse-scale mesh design divides the entire landslide area into multiple coarse-scale cells. This coarse-scale mesh is primarily used to describe the large-scale stress transfer effects of the landslide and conduct preliminary stability assessments. In relatively flat areas of the landslide or where stress variations are less pronounced, a coarse-scale mesh is used to reduce computational resource consumption. The coarse-scale mesh is primarily used to calculate the overall safety factor of the landslide, stability analysis, and potential sliding surface of the landslide using the transfer coefficient method.

[0121] Locally refined mesh design utilizes finer mesh elements in key areas of the landslide, such as the sliding surface, boundary layer, and areas of concentrated deformation. Using the material point method, the refined mesh constructs a three-dimensional material point model for mechanical simulation, capturing the nonlinear deformation of the landslide in these localized areas, particularly the potential sliding surface and fracture zones, ensuring high-precision simulation of the nonlinear failure process. This locally refined mesh accurately describes the complex stress distribution and large deformation processes within the landslide, particularly the stress and displacement fields at the sliding surface or potential failure zone.

[0122] The grid size setting can be roughly based on the fine grid size of 1m×1m and the coarse grid size of 10m×10m. At the same time, the following solutions can be used for cross-scale data switching:

[0123] Coarse→fine: Bicubic spline interpolation method is used to ensure the continuity of stress field;

[0124] Fine→Coarse: Gradient weighted average method is used. The specific formula is:

[0125] ;

[0126] ;

[0127] Where, represents the average stress value of the coarse-scale grid element; Indicates the The weight coefficient of each fine grid cell; Indicates the The stress value of each fine mesh element; Indicates the The stress gradient tensor for each fine mesh element.

[0128] The multi-scale collaborative simulation strategy, by organically combining coarse and fine-scale grids, not only ensures high-precision simulation results but also effectively reduces computational costs. The coarse-scale grid is suitable for large-scale stability analysis, while the fine-scale grid focuses on the precise calculation of local deformation, avoiding the enormous computational complexity associated with traditional methods of high-precision simulation of the entire landslide.

[0129] Furthermore, the present invention also provides a real-time dynamic feedback mechanism, which specifically includes: using a sensor network to collect real-time monitoring data of the landslide body, including but not limited to rainfall, landslide displacement, surface stress and pore water pressure; using dynamic data fusion technology to combine real-time monitoring data with historical landslide data, and dynamically adjust the model parameters in the material point method and transfer coefficient method; during the monitoring process of the landslide body, when the monitoring data changes significantly, retrain the model based on the machine learning algorithm and adjust the prediction results to ensure that the model can quickly respond to the dynamic changes of the landslide body.

[0130] By incorporating real-time sensor data and enabling real-time updates through data fusion and machine learning algorithms, the model can dynamically adjust its assessment results as the environment changes, improving its ability to predict the long-term stability of landslides and its ability to respond immediately to sudden landslide events. Dynamic adjustment and real-time updates of both models based on real-time monitoring data allow the models to respond to changes in landslides at different temporal and spatial scales.

[0131] Furthermore, the deformation fluctuation curve is used to characterize the monitoring data and evaluate the error. for:

[0132] ;

[0133] Where, Dynamic safety factor The standard deviation of .

[0134] Specifically, a real-time dynamic feedback mechanism includes:

[0135] A. Dynamic safety warning: by calculating the real-time stability coefficient , combined with physical perception to generate adversarial network model to optimize parameter calculation , the landslide body is dynamically graded for safety as shown in the table below, and early warning information of dangerous areas is output.

[0136]

[0137] B. Adaptive correction: Observe the deformation fluctuation curve When the monitoring data fluctuates or the model assessment error is too large, the model parameters are adjusted through the real-time correction mechanism to ensure the accuracy and reliability of the landslide assessment. Based on the real-time monitoring data and the prediction error of the model, the gradient descent method is used to adjust the parameters:

[0138] ;

[0139] ;

[0140] ;

[0141] Where, represents the real-time loss function for landslide assessment; 、 represents the weight, , ; Represents the learning rate, its initial value ; Representation parameters, such as friction coefficient, cohesion, etc.; Used to prevent the denominator from being 0, .

[0142] C. According to the warning level, immediately implement response measures, evacuate the crowd urgently, and carry out local or overall reinforcement after the sliding deformation stabilizes.

[0143] D. Automated landslide prediction: Use real-time feedback data to adjust the model to predict the medium- and long-term trends of landslide bodies, providing the time range for landslide instability and the possible danger level.

[0144] See also Figure 3 The present invention further provides a landslide stability assessment system combining a material point method with a transfer coefficient method. The system is applied to the landslide stability assessment method combining a material point method with a transfer coefficient method described above. The system comprises:

[0145] A 3D material point model construction module is used to construct a 3D material point model of the landslide based on the material point method to obtain the stress field distribution, deformation characteristics, and fluid-solid coupling effects of the landslide;

[0146] The real-time stability coefficient calculation module is used to divide the landslide into multiple blocks and determine the block area and position of each material point. It constructs a stress transfer matrix with the position and physical information of the material points as elements, and calculates the normal stress in the vertical direction between the blocks of the landslide and the tangential stress parallel to the contact surface between the blocks based on the stress transfer matrix. The real-time stability coefficient is calculated based on the contact force calculated from the normal stress and tangential stress.

[0147] The stability assessment module is used to calculate the dynamic safety factor of the landslide based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability coefficient obtained from the transfer coefficient method, and to perform stability assessment on the landslide based on the dynamic safety factor.

[0148] See also Figure 4 ,The present invention also provides a landslide stability assessment device combining a material point method and a transfer coefficient method, comprising a memory and a processor;

[0149] The memory is configured to store computer program code and transmit the computer program code to the processor;

[0150] The processor is configured to execute the landslide stability assessment method combining the material point method with the transfer coefficient method as described above according to the instructions in the computer program code.

[0151] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the landslide stability assessment method combining the material point method with the transfer coefficient method as described above is implemented.

[0152] Generally speaking, computer instructions for implementing the method of the present invention may be carried by any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media except for signals that are temporarily propagating.

[0153] Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or device.

[0154] Computer program code for performing the operations of the present invention can be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. In particular, Python, which is suitable for neural network computing, and platform frameworks such as TensorFlow and PyTorch can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or to an external computer (for example, through the Internet using an Internet service provider).

[0155] The above-mentioned device and non-transitory computer-readable storage medium can be referred to the detailed description of a landslide stability assessment method combining a material point method with a transfer coefficient method and its beneficial effects, which will not be repeated here.

[0156] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A landslide stability assessment method combining the material point method with the transfer coefficient method, characterized in that: include: Based on the material point method, a three-dimensional material point model of the landslide body is constructed to obtain the stress field distribution, deformation characteristics and fluid-solid coupling effect of the landslide body; Divide the landslide body into multiple block areas and determine the block areas and positions of each material point; A stress transfer matrix is constructed with the location and physical information of material points as elements. Based on the stress transfer matrix, the normal stress in the vertical direction between each block area of the landslide body and the tangential stress parallel to the contact surface between each block area are calculated. The real-time stability coefficient is calculated using the contact force calculated from the normal stress and tangential stress. The construction of the stress transfer matrix using the positions and physical information of material points as elements includes: ； ； ； ； ； ； ； ； ； ； ； ； Where, Represents the stress information matrix with the position and physical information of material points as elements; Indicates the position of a material point; Indicates the The coordinates of the points; Represents the physical information of a material point; Indicates the Physical information of each point; Represents the stress information matrix after transmission; Indicates the position of the material point after transfer; Represents the physical information of the material point after transfer; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; Indicates the The first block area points Coordinate value; 、 、 Respectively represent The first block area The density, volume, and stress information carried by each material point; The calculation formulas for the normal stress in the vertical direction between the block areas of the landslide body and the tangential stress parallel to the contact surface between the block areas are as follows: ； ； ； ； ； ； Where, represents the stress tensor; 、 、 Respectively 、 、 principal stresses in the direction; 、 、 Respectively 、 、 Tangential stress in the direction; represents the boundary normal vector; 、 、 Respectively represent the components of the boundary normal vector in each direction; represents the normal component; represents the tangential component; represents the transpose of the boundary normal vector; It represents the normal stress in the vertical direction between the blocks of the landslide body; It represents the tangential stress parallel to the contact surface between the various blocks of the landslide body; Indicates the contact area of each block of the landslide body; The dynamic safety factor of the landslide is calculated based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability factor obtained from the transfer coefficient method. The stability of the landslide is then evaluated based on the dynamic safety factor. The dynamic safety factor is: ； Where, express Dynamic safety factor at the moment; express Real-time stability coefficient at each moment; Indicates the influence coefficient of the real-time stability coefficient; express displacement of moments; Indicates a time interval; Indicates the influence coefficient of deformation rate; express Pore water pressure at the moment; Indicates the influence coefficient of pore water pressure; express The stress gradient norm at time t; Represents the influence coefficient of stress gradient.

2. The landslide stability assessment method combining the material point method and the transfer coefficient method according to claim 1 is characterized in that: Reconstruct historical landslide data based on the generative adversarial network model, including: A physics-aware generative adversarial network model was constructed, with a geomechanical equation solver embedded in the generator to generate landslide data that conforms to physical constraints. A dual-channel discriminator was designed, with the first channel assessing the authenticity of the generated landslide data and the second channel verifying the physical consistency of the stress field divergence and pore water pressure gradient. A hierarchical physical loss function is constructed, which includes global momentum conservation constraints and local contact surface stress distribution constraints. The weights of the hierarchical physical loss function are dynamically adjusted according to the real-time monitoring data of the landslide body, resulting in an optimized physical-aware generative adversarial network model. Historical landslide data are reconstructed based on the optimized physical-aware generative adversarial network model.

3. The landslide stability assessment method combining the material point method and the transfer coefficient method according to claim 2 is characterized in that: The layered physical loss function is: ； ； Where, represents the global physical loss function; represents the weight of the error term in the equilibrium equation; represents the divergence of the stress tensor; represents the stress tensor used to generate the landslide data; Indicates density; represents the acceleration due to gravity; represents the weight of the error term of the shear failure criterion; represents the shear stress tensor used to generate the landslide data; Indicates cohesion; represents the normal component; represents the internal friction angle; represents the local boundary physical loss function; represents the weight of the local stress rationality item; represents the boundary normal vector; Represents the weight of the stress tensor symmetry and projection error term; The method of dynamically adjusting the weight of the layered physical loss function according to the real-time monitoring data of the landslide body includes: ； ； ； ； ； Where, represents the momentum weight decay rate; represents the deformation rate; represents the pore water pressure gradient; represents the rate of change of the boundary normal vector; It represents the sensitivity coefficient of stress symmetry of complex terrain; represents the gradient of the sliding surface direction field; 、 、 、 are the initial values of each weight.

4. The landslide stability assessment method combining the material point method and the transfer coefficient method according to claim 1 is characterized in that: For loose soil landslides, the area with a stress gradient greater than or equal to 15 kPa or a displacement deformation rate greater than or equal to 2 mm / h is divided into a local complex refinement area; For rock landslides, the area with a stress gradient greater than or equal to 20 kPa or a displacement deformation rate greater than or equal to 1.5 mm / h is divided into a local complex refinement zone; For local complex and refined areas, a three-dimensional material point model of the landslide body is constructed based on the material point method.

5. A landslide stability assessment system combining the material point method and the transfer coefficient method, characterized in that: The system is applied to the method according to any one of claims 1 to 4, and the system comprises: A 3D material point model construction module is used to construct a 3D material point model of the landslide based on the material point method to obtain the stress field distribution, deformation characteristics, and fluid-solid coupling effects of the landslide; The real-time stability coefficient calculation module is used to divide the landslide into multiple blocks and determine the block area and position of each material point. It constructs a stress transfer matrix with the position and physical information of the material points as elements, and calculates the normal stress in the vertical direction between the blocks of the landslide and the tangential stress parallel to the contact surface between the blocks based on the stress transfer matrix. The real-time stability coefficient is calculated based on the contact force calculated from the normal stress and tangential stress. The stability assessment module is used to calculate the dynamic safety factor of the landslide based on the displacement, pore water pressure and stress obtained from the three-dimensional material point model and the real-time stability coefficient obtained from the transfer coefficient method, and to perform stability assessment on the landslide based on the dynamic safety factor.

6. A landslide stability assessment device combining the material point method and the transfer coefficient method, characterized in that: including memory and processor; The memory is configured to store computer program code and transmit the computer program code to the processor; The processor is configured to execute the method according to any one of claims 1 to 4 according to instructions in the computer program code.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which implements the method according to any one of claims 1 to 4 when executed by a processor.

Images