Slope stability prediction method based on deep learning

Abstract

The invention belongs to the technical field of measurement and data processing, and particularly relates to a slope stability prediction method based on deep learning, and the method comprises the steps: 1, carrying out the periodic scanning of the surface of a target slope through a high-precision laser radar, and obtaining point cloud data; 2, according to the geological data and hydro meteorological data of the slope area corresponding to each rock mass unit, the expansion potential energy of the slope area corresponding to each rock mass unit at each time in the water absorption process is calculated; 3, inputting the mean value output as an input quantity into a pre-trained deep learning auto-encoder, and outputting a potential variable in combination with variance output; and calculating a difference value between the potential variable and the mean value output, and carrying out ratio budgeting on the difference value and a preset basic threshold value to obtain the instability probability of the slope area corresponding to each rock mass unit. The method can dynamically respond to environmental changes, provides a stability evaluation result in real time, and has high prediction precision, real-time performance and adaptability.

Description

Technical Field

[0001] The present invention belongs to the technical field of measurement and data processing, and specifically relates to a slope stability prediction method based on deep learning. Background Art

[0002] With the development of society and the increasing human activities, slope instability has become an important safety hazard in mountainous areas, mining areas, highway construction and other fields. In particular, due to its unique physical and chemical properties, swelling rock slopes are prone to deformation and instability under the influence of natural and human factors, posing a great threat to engineering construction and the safety of people's lives and property. Therefore, how to accurately predict the stability of swelling rock slopes has become a technical problem facing the current geotechnical engineering field. The stability of swelling rock slopes is affected by many factors, mainly including geological characteristics, hydrological and meteorological conditions, geotechnical engineering properties and human activities. Swelling rock, as a rock containing swelling minerals such as montmorillonite, undergoes significant changes in rock volume during water absorption and swelling, resulting in reduced rock strength and deformation. In this process, factors such as water penetration rate, precipitation intensity, mineral composition of the rock mass and slope geometry all have an important impact on the stability of swelling rock. Therefore, traditional slope stability prediction methods usually rely only on experience or simple physical models, and often fail to fully consider the comprehensive impact of these complex factors.

[0003] At present, the research on the prediction of the stability of swelling rock slopes mainly focuses on traditional mechanical models and empirical data analysis methods. Common slope stability analysis methods include limit equilibrium method, finite element method, numerical simulation method, etc. The limit equilibrium method analyzes the stability of the slope by assuming a sliding surface, which is suitable for situations where the calculation is simplified and the slope conditions are relatively simple. However, this method often ignores the nonlinear and heterogeneous properties of swelling rock, and cannot effectively deal with complex hydrological and meteorological factors, and has poor accuracy and adaptability. The finite element method and numerical simulation method can better consider the deformation characteristics of swelling rock and perform numerical calculations by establishing a three-dimensional model of the slope. These methods have advantages in dealing with the geometric morphology of the slope and the anisotropy of the rock mass, and can more accurately simulate the deformation of swelling rock under external loads and hydrological environment. However, the finite element method and numerical simulation method require a lot of computing resources and have high requirements for input parameters. Especially under complex hydrological and meteorological conditions, how to accurately model the penetration and distribution of water is still a technical problem. In addition, the calculation results of these methods often rely on expert experience and require detailed field survey data support, which increases the complexity of calculation and the difficulty of data collection. In addition, some slope stability prediction models based on statistical methods and machine learning have also received widespread attention in recent years. Statistical methods establish empirical models by analyzing historical data, and combine regression analysis, cluster analysis and other techniques to mine the potential laws of slope instability from the data. This type of method has good data adaptability and can avoid the assumption bias in traditional mechanical models to a certain extent. However, statistical methods rely on a large amount of historical data and are often limited in dealing with nonlinear relationships. Especially for complex geotechnical media such as expansive rock, the prediction accuracy of statistical methods is low, and it is difficult to capture the dynamic changes under the coupling of multiple factors. Summary of the invention

[0004] In view of this, the main purpose of the present invention is to provide a slope stability prediction method based on deep learning, which can dynamically respond to environmental changes and provide stability assessment results in real time. It has high prediction accuracy, real-time and adaptability, overcomes the limitations of traditional methods in dealing with nonlinearity, multi-factor coupling and data scarcity, and provides a scientific basis for slope monitoring and disaster warning.

[0005] The technical solution adopted by the present invention is as follows: A slope stability prediction method based on deep learning, the method comprising: Step 1: Periodically scan the target slope surface with a high-precision laser radar to obtain point cloud data; segment the point cloud data to obtain multiple rock mass units; perform the following operations in the slope area corresponding to each rock mass unit: obtain the geological data of the slope area through geological surveys, and obtain the hydrological and meteorological data of the slope area at multiple continuous times through meteorological station records; Step 2: According to the geological data and hydro-meteorological data of the slope area corresponding to each rock mass unit, the expansion potential energy of the slope area corresponding to each rock mass unit at each time during the water absorption process is calculated; according to the expansion potential energy, the volume strain and the principal strain under isotropic conditions of the slope area corresponding to each rock mass unit at each time are calculated, and according to the volume strain and the principal strain, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated, and then the mean of the effective stress at all times is calculated to obtain the mean output; then the variance of the effective stress at all times is calculated to obtain the variance output; Step 3: Input the mean output as input into the pre-trained deep learning autoencoder, combine it with the variance output, and output the latent variable; calculate the difference between the latent variable and the mean output, and compare the difference with the preset basic threshold to obtain the instability probability of the slope area corresponding to each rock mass unit.

[0006] Furthermore, the geological data of the slope area corresponding to each rock unit obtained in step 1 include: rock dry density, average rock and soil particle size, water content, expansion pressure, bulk density of water, rock and soil mineral hydration heat, rock elastic modulus, Poisson's ratio, cohesion, internal friction angle, montmorillonite content and quartz content; the rock elastic modulus is measured by collecting rock samples from each rock unit and performing a uniaxial compression test on the rock samples; the cohesion and internal friction angle are measured by collecting rock samples from each rock unit and performing a direct shear test on the rock samples; the Poisson's ratio represents the lateral deformation ratio in another direction when the rock sample is stretched or compressed in one direction.

[0007] Furthermore, the hydrological and meteorological data of the slope area corresponding to each rock mass unit obtained in step 1 include: infiltration rate, rainfall and temperature.

[0008] Furthermore, in step 1, the surface of the target slope is periodically scanned by a high-precision laser radar to obtain point cloud data; the method of segmenting the point cloud data to obtain multiple rock mass units includes: using the following formula, Adaptive filtering is performed on each point: ; in, For point Point cloud density at ; To search for points within the sphere; is the radius of the search sphere; For point The distance to the nearest neighbor; is the standard deviation of the Gaussian distribution; For point The normal vector at ; is the laser radar line of sight direction vector; then use the following formula to determine the point Neighbors Is it a rock mass unit? ; in, is the judgment value, if , then the judgment point Its neighboring points in a rock mass unit; For point Its neighboring points The Euclidean distance of For point Neighbors The normal vector at ; For point Point cloud density at ; is the set angle threshold; is the set distance threshold; is the density threshold set.

[0009] Furthermore, in step 2, the expansion potential energy of each rock unit during the water absorption process at each time is calculated according to the geological data and hydrological and meteorological data of each rock unit by the following formula: ; in, For the The rock mass unit The expansion potential energy during the water absorption process over time; For the The dry density of rock mass in the slope area corresponding to each rock mass unit; For the The average particle size of rock and soil in the slope area corresponding to each rock mass unit; For the The water content of the slope area corresponding to each rock mass unit; For the The expansion pressure of the slope area corresponding to each rock mass unit; For the The bulk density of water in the slope area corresponding to each rock mass unit; For the The hydration heat of rock and soil minerals in the slope area corresponding to each rock mass unit; For the The average height of the slope area corresponding to each rock mass unit; For the The rock mass elastic modulus of the slope area corresponding to each rock mass unit; For the Poisson's ratio of the slope area corresponding to each rock mass unit; For the The cohesion of the slope area corresponding to each rock mass unit; For the The internal friction angle of the slope area corresponding to each rock mass unit; For the The montmorillonite content of the slope area corresponding to each rock mass unit; For the Quartz content of the slope area corresponding to each rock mass unit; is the gas constant; is the absolute temperature; for The temperature of time; for rainfall over time; for The penetration rate over time.

[0010] Furthermore, in step 2, the volume strain of the slope area corresponding to each rock mass unit at each time and the principal strain under isotropic conditions are calculated according to the expansion potential energy by the following formula: ; in, For the The volume of a rock mass unit; is the volume strain; under the isotropic conditions, the three principal strains are equal, and the calculation formula is: ; in, , and Take the initiative to respond.

[0011] Furthermore, in step 2, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated by the following formula: ; in, For the The slope area corresponding to the rock mass unit is Effective stress of time.

[0012] Furthermore, in step 3, the mean output is input as the input to the pre-trained deep learning autoencoder, and the process of outputting the latent variable specifically includes: assuming that the mean output is ; The mean output is encoded using the following formula: ; in, is the encoding result; is the mean distribution function; is the logarithmic variance distribution function; ;in, is the number of characteristic expansions in the past, and its value is equal to the target slope as the center, with the setting The value is the radius, the number of landslides that occurred in the circular area; ; The variance of the effective stress at all times is ; represents a normal distribution with a mean of 0 and a variance of 1; the encoded result is then input into a single-layer deep learning network to output the latent variable ; is the variance output; It is Hadamard.

[0013] Furthermore, in step 3, the latent variables are calculated by the following formula: ; in, Representation calculation The maximum value of Represents the activation function.

[0014] By adopting the above technical solution, the present invention produces the following beneficial effects: The present invention successfully overcomes the limitations of traditional methods by integrating physical modeling with deep learning technology. Traditional slope stability prediction methods, such as the limit equilibrium method and the finite element method, usually rely on simplified assumptions and cannot fully consider the nonlinear and spatiotemporal variation characteristics of expansive rocks under complex hydrological and meteorological conditions. These methods are often difficult to meet the needs of high precision in practical applications, especially when dealing with dynamic changes, complex geotechnical media and the coupling of multiple factors. In contrast, the present invention achieves a more comprehensive and dynamic stability prediction by combining physical quantities such as expansion potential energy, volumetric strain and effective stress with a deep learning model. The deep learning model can automatically extract key features in expansive rock slopes through an autoencoder without relying too much on manually set assumptions, making the prediction model more flexible, accurate and adaptable.

[0015] The present invention significantly improves the adaptability of the model to complex environments by introducing dynamic hydrometeorological data and historical landslide information. The stability of swelling rock slopes is affected by many factors, especially hydrometeorological conditions such as precipitation, temperature, and infiltration rate have a direct impact on the water absorption and expansion characteristics of swelling rocks. However, existing methods are usually unable to effectively process multi-dimensional hydrometeorological data, and ignore the long-term impact of historical landslide events on the stability of current slopes. By introducing deep learning-based coding and latent variable modeling, the present invention can integrate multi-source hydrometeorological data and historical landslide times in real time, thereby improving the comprehensive assessment capability of slope stability. For example, by encoding the effective stress and historical landslide times of each rock unit, the model can capture the potential impact of historical landslide events on the stability of the current slope and dynamically adjust the prediction results. This comprehensive modeling based on historical events and dynamic hydrological conditions makes the prediction results more consistent with the changing laws of actual slope stability, and has stronger accuracy and reliability.

[0016] The deep learning autoencoder of the present invention improves the accuracy and efficiency of feature expression by efficiently compressing and abstracting the stability features of the swelling rock slope. By compressing the input data through the autoencoder, the present invention can convert multi-dimensional and complex data into a low-dimensional latent space representation and extract the most important feature information. This process not only reduces the complexity of data processing, but also improves the computational efficiency of the model when facing high-dimensional data. The deep learning model can adaptively adjust the weight of the model according to the characteristics of the input data, thereby realizing the precision and personalization of the prediction of the stability of the swelling rock slope. This feature compression and abstraction method enables the model to learn quickly in a complex environment, and to perform customized predictions for different types of slopes and different hydrological and meteorological conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a method flow of a slope stability prediction method based on deep learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] All features disclosed in this specification, or steps in all methods or processes disclosed, except mutually exclusive features and / or steps, can be combined in any manner.

[0019] Any feature disclosed in this specification (including any additional claims and abstract), unless otherwise stated, may be replaced by other alternative features that are equivalent or have similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

[0020] Example 1: Reference Figure 1, a slope stability prediction method based on deep learning, the method comprising: Step 1: Periodically scan the target slope surface with a high-precision laser radar to obtain point cloud data; segment the point cloud data to obtain multiple rock mass units; perform the following operations in the slope area corresponding to each rock mass unit: obtain the geological data of the slope area through geological surveys, and obtain the hydrological and meteorological data of the slope area at multiple continuous times through meteorological station records; Traditional manual measurement methods usually cannot meet the requirements of slope monitoring for accuracy and efficiency, while LiDAR can achieve high-density coverage scanning of the entire slope area in a short time, and its spatial accuracy can usually reach the centimeter level or even the millimeter level, which is particularly suitable for capturing the microscopic deformation characteristics of swelling rock during water absorption. Secondly, LiDAR adopts non-contact measurement, which not only ensures the safety of data acquisition, but also avoids disturbance to the slope surface, thereby maintaining the authenticity and reliability of the data. After obtaining the point cloud data, it must be segmented and processed to extract the rock units that reflect the characteristics of the specific area of ​​the slope. Point cloud segmentation is to divide a large amount of original point cloud data into several subsets with geometric consistency or specific attributes according to certain rules. In the present invention, this process is usually implemented by algorithms based on geometric features, such as regional growing methods, density clustering algorithms (such as DBSCAN), etc. These algorithms can classify points with similar characteristics according to the characteristics of the distance, normal vector, etc. of the points, thereby forming sub-areas representing specific rock units. The division of rock mass units provides a structured basis for subsequent data analysis, so that complex point cloud data can be gradually simplified into discrete calculation units.

[0021] Once the point cloud segmentation is completed, each rock unit can be given rich physical meaning in combination with geological survey and meteorological monitoring data. Geological survey data usually includes information such as rock type, distribution of cracks, and mineral composition, which directly affect the water absorption and expansion characteristics of the rock mass. For example, different mineral compositions may lead to significant differences in the expansion rate of rocks after absorbing water, while the distribution of cracks affects the diffusion path of water inside the rock mass. In addition, the continuous time hydrometeorological data (such as rainfall, air temperature and humidity) provided by the meteorological station provides support for the dynamic monitoring of the environmental conditions of the swelling rock. These data are combined to establish a complete characteristic description for each rock unit. The significance of step 1 is to organically integrate multi-source data and dynamically update them at different time points to provide high-quality input for the entire model. In traditional slope stability analysis, data acquisition often has problems such as discontinuous collection, insufficient accuracy or limited coverage, resulting in large errors in the description of rock behavior. Through high-precision lidar and point cloud segmentation technology, the present invention can achieve refined capture of the geometric morphology and structural characteristics of the slope surface, and at the same time, through the auxiliary supplement of geological and meteorological data, integrate dynamic environmental conditions into the characteristic modeling of rock units. In this way, each rock mass unit not only represents its geometric properties, but also carries multidimensional physical and environmental information, so that the subsequent calculation of expansion potential and stress state can accurately reflect its true behavior.

[0022] Step 2: According to the geological data and hydro-meteorological data of the slope area corresponding to each rock mass unit, the expansion potential energy of the slope area corresponding to each rock mass unit at each time during the water absorption process is calculated; according to the expansion potential energy, the volume strain and the principal strain under isotropic conditions of the slope area corresponding to each rock mass unit at each time are calculated, and according to the volume strain and the principal strain, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated, and then the mean of the effective stress at all times is calculated to obtain the mean output; then the variance of the effective stress at all times is calculated to obtain the variance output; The expansion characteristics of swelling rock mainly come from hydrophilic minerals such as montmorillonite in its internal mineral structure. When water molecules enter the lattice interlayers of these minerals, the lattice will expand and produce potential energy changes. This expansion potential is the fundamental reason for driving the volume change and deformation of the rock unit. In the present invention, the calculation of the expansion potential fully considers the mineral composition, water absorption capacity and fracture structure of the rock mass, as well as the dynamic changes of meteorological conditions. By introducing multi-time-step hydrometeorological data, the nonlinear behavior of water entering the rock mass can be accurately simulated. This high-timeliness and high-resolution calculation provides an accurate basis for subsequent stress state analysis. On the basis of the calculation of the expansion potential, the volume strain and principal strain of the rock unit are further derived. These parameters are important physical quantities that describe the deformation characteristics of the rock mass. The volume strain reflects the total volume change of the rock mass during the water absorption process, which is directly related to the expansion potential, while the principal strain further describes the deformation components of the rock mass in different directions through the isotropic assumption. After swelling rock absorbs water, its volume change and directional deformation have a significant impact on slope stability. This deformation not only changes the geometric shape of the rock mass, but also affects the redistribution of internal stress. By combining actual geological data, the present invention refines the calculation of volume strain and principal strain so that it can truly reflect the behavioral characteristics of swelling rock.

[0023] After calculating the volume strain and the principal strain, the present invention directly links the deformation characteristics of the swelling rock with its stability state through the calculation of the effective stress. Effective stress is a key parameter in the rock mass stability analysis, which reflects the actual stress state of the rock mass under the combined action of the external load and the internal pore water pressure. During the water absorption process of the swelling rock, the effective stress will be significantly reduced due to the gradual increase of water in the pores, thereby increasing the risk of slope instability. The calculation of the effective stress of the present invention fully considers the dynamic change of the pore water pressure, and obtains the effective stress distribution of the rock unit at different times through the calculation of multiple time steps. Such dynamic calculation results not only improve the understanding of the law of change of the stability of the swelling rock slope, but also provide a scientific basis for the quantitative analysis of the probability of instability. After completing the calculation of the effective stress at all times, the present invention further extracts the mean and variance of the effective stress as feature outputs. The mean of the effective stress can reflect the average stress state of the rock mass over the entire time range, while the variance reveals the degree of fluctuation of the effective stress. The combination of these two parameters can fully reflect the stability of the swelling rock slope, because when the mean is low, the slope is more likely to enter an unstable state, while a large variance may indicate that the rock mass is more unstable and there is a potential risk of damage. This method of extracting the mean and variance not only reveals the dynamic behavior of swelling rock from a physical level, but also provides high-quality input features for the deep learning model.

[0024] Step 3: Input the mean output as input into the pre-trained deep learning autoencoder, combine it with the variance output, and output the latent variable; calculate the difference between the latent variable and the mean output, and compare the difference with the preset basic threshold to obtain the instability probability of the slope area corresponding to each rock mass unit.

[0025] An autoencoder is a neural network model for unsupervised learning. Its basic principle is to compress high-dimensional data into a low-dimensional latent space through an encoder, and then reconstruct it into the original data through a decoder. This structure can effectively extract the core features of the data and show strong modeling capabilities when the data has high-dimensional nonlinear relationships. In the present invention, the autoencoder is used to process the mean and variance of the effective stress of the expansive rock slope. These two feature quantities contain the stability information of the rock unit under multi-time step dynamic loading. The mean of the effective stress reflects the overall stress level of the rock unit, while the variance reveals the magnitude of its stress fluctuation, which may be caused by changes in meteorological conditions, differences in rock mass properties or external loads. By inputting the autoencoder, this information is further compressed and mapped to the latent space to generate latent variables that can reflect complex feature relationships. The generation process of latent variables reflects the powerful ability of deep learning in dealing with nonlinear problems. In traditional methods, feature engineering often relies on artificially set rules and assumptions, and it is difficult to capture the potential high-order relationships between data. The autoencoder automatically learns the implicit relationship between features through a neural network, which can not only retain the physical meaning of the effective stress mean and variance, but also mine the nonlinear patterns hidden behind these feature quantities. For example, in an expansive rock slope, rainfall intensity may affect the stress distribution of the rock mass through complex water infiltration paths, and these complex coupling effects cannot be directly described by linear models. The autoencoder extracts and retains these nonlinear relationships through the representation of the latent space, providing a comprehensive feature description for the subsequent calculation of the instability probability.

[0026] The generation of latent variables does not directly reflect the stability of the slope, but needs to be further compared with the effective stress mean to reveal the instability risk of the swelling rock slope. The present invention quantifies the probability of instability of each rock unit by calculating the difference between the latent variable and the mean output and normalizing it to a ratio related to a preset threshold. The difference here can be regarded as a deviation, reflecting the difference between the stress state of the current rock unit and the potential mode. If this difference is significant, it means that the stress state of the rock unit is abnormal and may be in a critical state of instability. Through normalization, this difference is further converted into a probability value, thereby realizing the quantitative characterization of the instability risk. It is worth noting that the preset basic threshold plays a key role in the whole step. The setting of the threshold is usually based on empirical data or engineering standards, reflecting the critical conditions for the rock unit to transition from a stable state to an unstable state. In swelling rock slopes, instability may be caused by local stress concentration or a sudden change in environmental conditions, so the selection of the threshold needs to comprehensively consider the physical properties of the swelling rock and the actual geological environment. In practical applications, the threshold can be adjusted to adapt to different slope environments and engineering requirements, thereby improving the adaptability and robustness of the model.

[0027] Example 2: The geological data of the slope area corresponding to each rock unit obtained in step 1 include: rock dry density, average rock and soil particle size, water content, expansion pressure, water density, rock and soil mineral hydration heat, rock elastic modulus, Poisson's ratio, cohesion, internal friction angle, montmorillonite content and quartz content; the rock elastic modulus is measured by collecting rock samples from each rock unit and performing a uniaxial compression test on the rock samples; the cohesion and internal friction angle are measured by collecting rock samples from each rock unit and performing a direct shear test on the rock samples; the Poisson's ratio represents the lateral deformation ratio in another direction when the rock sample is stretched or compressed in one direction.

[0028] Specifically, the dry density of rock mass is an important indicator reflecting the compactness of rock mass, which is defined as the mass of rock mass per unit volume in a dry state. The dry density is closely related to the porosity of the rock mass, which directly affects the water penetration rate and water absorption capacity inside the rock mass. In an expansive rock slope, a higher dry density usually means a lower porosity and water absorption capacity, which may slow down the expansion process, but may also cause local stress concentration, thereby increasing the risk of instability. The present invention provides basic parameters for subsequent expansion potential energy calculations by accurately measuring the dry density of the rock mass unit. The average particle size of rock and soil describes the geometric characteristics of rock mass particles and is an important parameter affecting the diffusion and expansion rate of water. The size of the particle size determines the distribution form of the permeability and hydraulic gradient of the rock mass inside. In an expansive rock slope, a smaller particle size may cause water to diffuse rapidly inside the rock mass, inducing faster expansion behavior. By obtaining the average particle size of rock and soil, the present invention can more accurately simulate the dynamic changes in water distribution during water absorption.

[0029] Water content is a basic indicator for describing the water content in rock mass, which directly determines the water absorption capacity and swelling degree of swelling rock. In the present invention, by real-time monitoring and recording the water content of each rock mass unit, the water absorption behavior of the rock mass can be dynamically tracked, and input can be provided for the calculation of swelling potential energy and volume strain. The change of this parameter can also reflect the influence of environmental conditions on the stability of the slope. For example, the rapid increase of water content caused by rainfall may be a precursor to the instability of the slope. Swelling pressure is an important indicator for describing the mechanical response of swelling rock due to volume expansion during water absorption. Swelling pressure is not only related to the physical properties of the rock mass, but also limited by external boundary conditions. In the present invention, the acquisition of swelling pressure provides a direct basis for subsequent stress calculation, and can also be used to evaluate the instability risk of rock mass units under water absorption. The bulk density of water is an important parameter affecting the pore water pressure in swelling rock, which is defined as the weight of water per unit volume. In swelling rock slopes, pore water pressure is the main factor leading to the decrease of effective stress, and effective stress is the core indicator of stability analysis. The present invention considers the bulk density of water and incorporates its effect into the expansion potential energy and stress calculation model, thereby more accurately reflecting the real mechanical state of the rock mass. The hydration heat of rock and soil minerals refers to the heat released or absorbed by the minerals in the rock mass when the hydration reaction occurs during the water absorption process. This parameter directly affects the water absorption rate and expansion potential of the swelling rock. Rock masses with a high content of hydrophilic minerals such as montmorillonite usually have a larger hydration heat, resulting in more violent expansion behavior during the water absorption process. The present invention provides a thermodynamic basis for the calculation of the expansion potential energy by measuring the hydration heat. The elastic modulus of the rock mass is an important parameter that describes the deformation resistance of the rock mass in the elastic deformation stage, which is determined by a uniaxial compression test. The uniaxial compression test places the rock mass sample under a stress state and measures the relationship between its deformation and stress to determine the elastic modulus. In the swelling rock slope, the elastic modulus determines the response speed and deformation degree of the rock mass to the swelling pressure. The present invention can better simulate the deformation characteristics of the rock mass during the expansion process by accurately measuring each rock mass unit. Poisson's ratio describes the ratio of lateral deformation to longitudinal deformation when a rock mass is subjected to force in one direction. This parameter is obtained through uniaxial compression test and is an important basis for the assumption of isotropy of rock mass. In the present invention, Poisson's ratio is used to calculate volume strain and principal strain, and is an important bridge for converting expansion potential energy into strain state. Cohesion and internal friction angle are two basic parameters reflecting the shear strength of rock mass, which are determined by direct shear test. Cohesion describes the cohesive force between rock mass particles, while the internal friction angle reflects the frictional resistance between rock mass particles. In the swelling rock slope, the water absorption process may reduce the cohesion and internal friction angle, thereby weakening the shear strength of the rock mass. The present invention provides a direct basis for the prediction of the probability of instability by accurately measuring these parameters. The content of montmorillonite and quartz is the key mineralogical parameter affecting the water absorption behavior of swelling rock.Montmorillonite has a strong water absorption and expansion property. The higher its content, the greater the expansion potential of the rock mass. Quartz has a high rigidity and has a certain inhibitory effect on the expansion behavior. In the present invention, by measuring the content of these two minerals, the expansion behavior of the rock mass during water absorption and its influence on the slope stability can be more accurately predicted.

[0030] Embodiment 3: The hydrological and meteorological data of the slope area corresponding to each rock mass unit obtained in step 1 include: infiltration rate, rainfall and temperature.

[0031] Specifically, the infiltration rate describes the rate at which water enters the rock mass from the surface, and is a direct factor affecting the water absorption behavior of expansive rock and the change in pore water pressure. The water absorption and expansion characteristics of expansive rock are highly sensitive to the infiltration rate. When the infiltration rate is high, a large amount of water quickly penetrates into the rock mass, causing the pore water pressure to increase rapidly, and further reducing the effective stress, which may accelerate the occurrence of slope instability. The size of the infiltration rate depends on factors such as the permeability of the rock mass, the distribution of cracks, and the intensity of rainfall. In expansive rocks, the degree of development of cracks often makes the spatial distribution of the infiltration rate highly heterogeneous. The present invention can more accurately reflect the real-time process of water entering the rock mass by dynamically monitoring and recording the infiltration rate of each rock unit, and provide a basic input for the calculation of expansion potential energy. Precipitation is one of the most direct meteorological parameters in the stability analysis of expansive rock slopes, and its influencing mechanism is mainly reflected in two aspects. On the one hand, rainfall directly determines the total amount of water supply, affecting the saturation state of the rock mass and the change in pore water pressure; on the other hand, heavy rainfall events may cause high-intensity infiltration in a short period of time, thereby inducing stress concentration and instability of local rock units. The change in rainfall is not only dynamic in time, but also shows differences in regional distribution. For example, changes in local rainfall intensity may lead to significant differences in water absorption rate and expansion degree of different rock units. The present invention obtains rainfall data of the slope area corresponding to each rock unit in real time, and combines geological and hydrological conditions to accurately model the expansion effect caused by rainfall, thereby improving the ability to predict the risk of slope instability. Temperature is an indirect factor affecting the water absorption and expansion process of expansive rocks, which mainly plays a role by changing the physical and chemical properties of water and the water absorption behavior of rock mass. Higher temperatures usually reduce the viscosity of water, increase the permeability rate, and thus accelerate the process of water entering the rock mass; at the same time, temperature may also affect the hydration reaction rate of rock minerals. For example, the water absorption and expansion behavior of expansive minerals such as montmorillonite may be more intense in a high temperature environment. In addition, the temperature changes during the day and night will cause the thermal expansion and contraction effect of the rock mass, which, together with the water absorption expansion, may lead to stress concentration and crack expansion in local areas of the slope. The present invention records the temperature data of each rock mass unit and incorporates it into the dynamic analysis of expansion potential and stress state to more comprehensively capture the impact of environmental conditions on slope stability.

[0032] Embodiment 4: In step 1, the surface of the target slope is periodically scanned by a high-precision laser radar to obtain point cloud data; the method of segmenting the point cloud data to obtain multiple rock mass units includes: using the following formula, Adaptive filtering is performed on each point: ; in, For point Point cloud density at ; To search for points inside the sphere; is the radius of the search sphere; For point The distance to the nearest neighbor; is the standard deviation of the Gaussian distribution; For point The normal vector at ; is the laser radar line of sight direction vector; then use the following formula to determine the point Neighbors Is it a rock mass unit? ; in, is the judgment value, if , then the judgment point Its neighboring points in a rock mass unit; For point Its neighboring points The Euclidean distance of For point Neighbors The normal vector at ; For point Point cloud density at ; is the set angle threshold; is the set distance threshold; is the density threshold set.

[0033] Specifically, the core of point cloud data adaptive filtering is to calculate the local density of each point , the density value not only reflects the point The density in space also takes into account its geometric characteristics and its relative relationship with the direction of the LiDAR line of sight. The first part of the density calculation formula searches for the number of points inside the sphere and the volume of the sphere Determine the basic density, this part directly captures the point The distribution characteristics of the surrounding local space. However, this density calculation may be limited on complex slope surfaces, because the geometric shape of swelling rocks is usually highly irregular, and the point cloud density that relies solely on geometric distribution may not fully reflect the local characteristics of the points. Therefore, the formula further introduces a Gaussian distribution factor , through the distance of the nearest neighbor Adjust the density value. Points that are closer contribute more to the density, while the influence of points that are farther away is significantly attenuated. This design not only improves the accuracy of point cloud density calculation, but also enhances the sensitivity to local geometric details. At the same time, the third part of the formula is calculated through the normal vector and the laser radar line of sight direction vector The inner product of the point cloud is used to further adjust the point cloud density. The physical basis of this design is that during the LiDAR measurement process, the angle between the normal vector direction and the line of sight direction will affect the intensity and accuracy of the laser reflection. When the normal vector is close to parallel to the line of sight, the density correction value is higher, which further emphasizes the importance of these points in the slope structure analysis. This density correction based on geometry and observation angle allows the formula to be adaptively adjusted, with higher resolution for complex areas on the slope surface (such as steep edges or convex parts).

[0034] After calculating the point cloud density, the formula further introduces the neighboring point judgment criterion To determine whether there is spatial, density and geometric continuity in the point cloud. The slope surface of the expansive rock often has a complex internal structure due to water swelling, crack development and other factors. This structural characteristic requires that the point cloud segmentation can meet multiple constraints at the same time. The first term of the neighbor point determination formula is the inner product of the normal vector With threshold The comparison ensures that the point clouds identified as the same rock mass unit have similar local surface geometric characteristics. Large differences in the directionality of the normal vectors indicate that there may be cracks or changes in material properties, so assigning these points to different units can better reflect the actual slope characteristics. In addition, the spatial distance between neighboring points Density difference are two other important criteria. Point cloud segmentation requires points and The Euclidean distance is less than the set threshold , this condition ensures the spatial continuity of the segmentation result. The density difference threshold It is used to further restrict the physical consistency of the points. Points with too large density differences usually reflect significant changes in material properties and may not belong to the same rock unit. This multi-condition comprehensive constraint enables the point cloud segmentation results to fully reflect the physical and geometric characteristics of the rock unit, providing accurate unit division for subsequent expansion potential energy calculation and stress analysis. The design of the formula fully considers the characteristics of the swelling rock slope and the observation characteristics of the laser radar scanning. Due to its water absorption and swelling characteristics, the swelling rock causes the slope surface to have high irregularity at the microscopic scale, and the changes in point cloud density, normal vector and observation angle during the laser radar scanning process further increase the complexity of data processing. By introducing a combination formula of adaptive filtering and multi-condition judgment, the present invention can effectively cope with these challenges, thereby ensuring that the rock unit can be accurately divided under different environmental conditions. The innovation of this point cloud segmentation method is not only reflected in the accuracy of its algorithm, but also in the seamless connection with the subsequent deep learning model. The segmented rock unit data has clear physical and geometric characteristics, which can be directly used as the basic input for the calculation of expansion potential energy and effective stress. The deep learning model further realizes the dynamic prediction of the probability of slope instability by learning the characteristic distribution in the segmentation results. Compared with traditional methods, this point cloud segmentation scheme based on a combination of formulas and deep learning not only improves the accuracy of slope stability prediction, but also significantly improves its adaptability to complex environments.

[0035] Embodiment 5: In step 2, the expansion potential energy of each rock unit during the water absorption process at each time is calculated according to the geological data and hydrological and meteorological data of each rock unit by the following formula: ; in, For the The rock mass unit The expansion potential energy during the water absorption process over time; For the The dry density of rock mass in the slope area corresponding to each rock mass unit; For the The average particle size of rock and soil in the slope area corresponding to each rock mass unit; For the The water content of the slope area corresponding to each rock mass unit; For the The expansion pressure of the slope area corresponding to each rock mass unit; For the The bulk density of water in the slope area corresponding to each rock mass unit; For the The hydration heat of rock and soil minerals in the slope area corresponding to each rock mass unit; For the The average height of the slope area corresponding to each rock mass unit; For the The rock mass elastic modulus of the slope area corresponding to each rock mass unit; For the Poisson's ratio of the slope area corresponding to each rock mass unit; For the The cohesion of the slope area corresponding to each rock mass unit; For the The internal friction angle of the slope area corresponding to each rock mass unit; For the The montmorillonite content of the slope area corresponding to each rock mass unit; For the Quartz content of the slope area corresponding to each rock mass unit; is the gas constant; is the absolute temperature; for The temperature of time; for rainfall over time; for The penetration rate over time.

[0036] Specifically, through the basic volume of rock and soil and dry density , calculate the mass distribution of rock mass units. Here, the average particle size of rock is It is a core geometric parameter in the swelling rock slope, which directly determines the porosity of the rock unit and the water diffusion path. The larger the particle size, the lower the porosity of the rock unit is usually, and the water absorption capacity may be limited, but the swelling pressure distribution inside the rock is more uniform. It indicates the mass distribution of the rock mass unit, and together with the particle size, determines the physical behavior of the rock mass during water absorption. By combining geometric properties with density properties, the formula can accurately describe the basic physical characteristics of the rock mass unit, which is the core basis for the calculation of expansion potential energy. and moisture content It is an important dynamic variable that reflects the process of rock unit absorbing water. The swelling pressure comes from the unique swelling phenomenon of swelling rock, that is, when water enters the pores or mineral lattices of the rock mass, it causes the volume increase effect inside the rock mass. It is a direct indicator to measure the water content inside the rock mass. The dynamic relationship between the change of swelling pressure and the water content is dynamically related. The higher the water content, the stronger the total effect of swelling pressure. This dynamic model can well capture the swelling behavior of rock units at different water absorption stages, especially under heavy rainfall or continuous infiltration conditions. When the water content rises rapidly, the swelling pressure will increase significantly, resulting in a rapid accumulation of swelling potential energy. In addition, the formula further considers the dynamic influence of hydrometeorological parameters on swelling potential energy, specifically through This term combines the infiltration rate , rainfall , water density and the average height of the rock mass unit . The infiltration rate and rainfall directly determine the amount of water that enters the rock mass per unit time, while the bulk density of water reflects the gravitational effect of water on the rock mass. The average height is related to the slope structure and the water infiltration path, which determines the distribution pattern of water under the action of gravity. The combination of these parameters enables the formula to simulate the driving effect of hydrological conditions on swelling potential energy on a macro scale. For example, in heavy rainfall events, high infiltration rates and high rainfall will significantly increase the water supply of rock units, further increase swelling pressure and water content, and thus trigger a rapid increase in swelling potential energy. This dynamic relationship is particularly critical in the instability process of swelling rock slopes, and can accurately capture the transient effects caused by hydrological conditions.

[0037] The formula introduces the thermodynamic term The core of this part is to describe the effect of hydration reaction of rock minerals on expansion potential. When rock minerals (such as montmorillonite) absorb water, water molecules enter the mineral lattice layers and hydration reaction occurs. This reaction is usually accompanied by the release or absorption of heat, and its degree is determined by the mineral hydration heat. Determine. The formula uses the thermodynamic model, through and absolute temperature The relationship between hydration reaction and expansion potential energy describes the dynamic regulation of hydration reaction on expansion potential energy. When the temperature is high, the rate and degree of hydration reaction may increase significantly, thereby accelerating the accumulation of expansion potential energy; while lower temperatures will inhibit this process. The introduction of this thermodynamic term enables the formula to adapt to the characteristics of expansive rocks under different climate and temperature conditions, especially in areas with significant day-night temperature differences or drastic seasonal changes, the impact of temperature fluctuations on expansion behavior can be accurately captured. The formula uses the elastic modulus and Poisson's ratio Combination of , further quantifies the deformation response of the rock mass unit under the action of expansion pressure. Elastic modulus It indicates the ability of rock mass to resist deformation within the elastic range, while Poisson's ratio It reflects the deformation ratio in one direction when the rock mass is compressed or stretched in another direction. It shows that the larger the Poisson's ratio, the lower the contribution of the lateral deformation of the rock mass to the expansion potential energy. This part of the formula closely combines the expansion pressure with the material properties of the rock mass, and can describe the expansion characteristics of the rock mass in different directions in detail. This refinement of the mechanical response provides basic data for the subsequent calculation of volumetric strain and principal strain, allowing the formula to more comprehensively reflect the expansion behavior of the rock mass unit. Finally, the last part of the formula is expressed by the friction angle , Cohesion , montmorillonite content and quartz content Parameters such as friction angle reveal the regulatory effect of mineral composition inside the rock mass on the expansion potential. and cohesion They are key indicators of rock mass shear strength. They not only affect the conversion efficiency of expansion potential energy, but are also directly related to the stability of the rock mass. The correction term describes the weakening effect of cohesion on the expansion pressure distribution. This relationship is particularly applicable to areas where there are obvious changes in cohesion inside the swelling rock. In addition, the montmorillonite content and quartz content The ratio of further reflects the influence of mineral composition on expansion behavior. Montmorillonite is the main expansion mineral of expansive rock. The higher its content, the greater the expansion potential of the rock mass; while quartz is a rigid mineral that can inhibit expansion behavior to a certain extent. The formula is obtained through the ratio Dynamically balancing the effects of these two minerals on the swelling potential, more accurately describes the role of the mineralogy of the rock mass unit in regulating the swelling process. This modeling of mineral composition is particularly important for the prediction of swelling rock slopes, because regional differences in mineral composition can lead to significant changes in swelling behavior, which is difficult to capture in traditional prediction models.

[0038] Embodiment 6: In step 2, the volume strain of the slope area corresponding to each rock mass unit at each time and the principal strain under isotropic conditions are calculated according to the expansion potential energy by the following formula: ; in, For the The volume of a rock mass unit; is the volume strain; under the isotropic conditions, the three principal strains are equal, and the calculation formula is: ; in, , and Take the initiative to respond.

[0039] Specifically, the volume strain is calculated based on the expansion potential energy The core of this energy conversion relationship is that the expansion potential energy is caused by the increase in internal energy caused by water entering the pores of the rock mass or the mineral lattice. The higher the expansion potential energy, the greater the expansion pressure on the rock mass, and the more significant its volume deformation. The formula is obtained through the elastic modulus. and Poisson's ratio The coupling of the elastic modulus and the elastic modulus of the rock mass clearly defines the constraint effect of the material properties of the rock mass during the expansion process. The elastic modulus describes the deformation resistance of the rock mass unit to the expansion pressure. The higher the value, the smaller the volume strain, indicating that the rock mass exhibits higher stiffness during the expansion process. The Poisson's ratio is calculated by the correction factor , which describes the effect of lateral deformation on volumetric strain. The lateral expansion effect is more significant for rock units with higher Poisson's ratio, thus reducing the accumulation of overall volumetric strain. It is an important scale parameter for the distribution of expansion potential energy inside the rock unit. The introduction of volume can adjust the contribution rate of expansion potential energy to volume strain, so that the formula can adapt to the deformation characteristics of rock units of different sizes. For example, for rock units with smaller volumes, the same expansion potential energy may induce more significant volume strain, while for rock units with larger volumes, the expansion potential energy will be more evenly dispersed, resulting in a weakening of the strain effect. This dynamic adjustment based on volume can truly reflect the deformation differences in different regions of the expansive rock slope and provide a more refined description for slope stability analysis. On the basis of volume strain, the formula further uses the isotropic assumption to divide the volume strain into three principal strains, defining the deformation components of the rock unit in each direction. This assumption shows that during the water absorption and expansion process, the deformation of the rock unit is uniformly distributed, and there is no obvious directional stress concentration. The calculation results of the principal strain not only reflect the overall deformation trend inside the rock mass, but also reveal the stability state of the local area. For example, when the principal strain value of a rock unit is significantly greater than that of the surrounding units, it may mean that there is an abnormal concentration of expansion pressure or insufficient material strength in the area, which may become a potential risk point for slope instability. Through the distribution calculation of the principal strain, the present invention further links the expansion potential energy with the deformation characteristics of the rock mass, providing key input for subsequent effective stress calculations and instability probability predictions. The magnitude of the principal strain directly affects the stress distribution inside the rock unit. When the principal strain is large, the stress concentration of the rock unit is also higher, which may lead to the risk of crack expansion, material damage and even overall instability. The uniformity of the principal strain reflects the coordination inside the rock mass, and a uniform distribution of principal strains usually means that the rock unit has good stability. This quantitative description of deformation characteristics not only improves the accuracy of slope stability analysis, but also provides richer feature data for deep learning models.

[0040] Embodiment 7: In step 2, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated by the following formula: ; in, For the The slope area corresponding to the rock mass unit is Effective stress of time.

[0041] Specifically, the first part of the formula focuses on the internal mechanical response of the expansive rock material, reflecting the stress distribution of the rock mass during the elastic deformation stage. and Poisson's ratio It is the core parameter of this part, which describes the resistance and deformation characteristics of rock materials to external forces. The higher the elastic modulus, the greater the stiffness of the rock mass, the smaller its deformation response, and the greater the internal stress generated; Poisson's ratio is expressed in its composite form. and , which relates volumetric strain to three-dimensional stress distribution, reflects the regulatory effect of lateral deformation on the total stress of the rock mass. In expansive rocks, the role of Poisson's ratio is particularly important because rock masses with significant lateral expansion may weaken stress concentration in some directions and enhance the inhomogeneity of stress distribution in other directions. This complex mechanical response is fully reflected in the formula, ensuring accurate modeling of the real stress state of the rock material. Volumetric strain It is the key variable that drives the generation of internal stress, and its value depends on the relationship between the expansion potential and the material properties of the rock mass. In expansive rock, water absorption and expansion directly lead to changes in the volume of the rock mass, and volume strain is a quantitative indicator of this change. When the expansion potential is high, the expansion pressure of the rock mass increases significantly, causing the volume strain to rise rapidly and the internal stress to increase accordingly. This dynamic relationship not only reflects the deformation mechanism of expansive rock, but also reflects the direct impact of environmental conditions on the mechanical behavior of the rock mass. For example, under heavy rainfall conditions, the internal stress of the expansion is concentrated and increased. The formula is expressed through volume strain The coupling with elastic modulus and Poisson's ratio accurately maps the energy conversion during the swelling process of expansive rock to the generation process of internal stress. This mapping not only covers the elastic properties of rock materials, but also captures the nonlinear changes under dynamic environmental conditions, making the model adaptable to complex hydrological and meteorological conditions. The second part of the formula focuses on the weakening effect of hydrological and meteorological conditions on the effective stress of rock units, which is specifically reflected by the pore water pressure term. Here, the specific gravity of water Describes the weight of water per unit volume, and its value reflects the basic contribution of water to pore pressure; the height of the rock unit The pressure effect of the water column is further defined. The greater the height, the higher the pore water pressure inside the rock mass. It is a dynamic variable that directly determines the variation of the water content inside the rock mass. When the water content increases, such as under heavy rainfall or continuous infiltration conditions, the pore water pressure increases significantly, thereby significantly weakening the effective stress of the rock mass. The introduction of pore water pressure is of particular importance in the stability analysis of expansive rock slopes. Expansive rocks often exhibit higher pore water pressures due to their high water absorption and significant volume expansion characteristics. This pressure not only reduces the ability of the rock mass to withstand external loads, but may also cause crack expansion inside the rock mass and weakening of material strength, ultimately leading to slope instability. Through the introduction of this term, the formula can accurately capture the impact of dynamic changes in water on the stress distribution of the rock mass, providing a reliable basis for the prediction of slope stability under complex hydrological conditions.

[0042] Example 8: In step 3, the mean output is input as an input to the pre-trained deep learning autoencoder, and the process of outputting the latent variable specifically includes: assuming that the mean output is ; The mean output is encoded using the following formula: ; in, is the encoding result; is the mean distribution function; is the logarithmic variance distribution function; ;in, is the number of characteristic expansions in the past, and its value is equal to the target slope as the center, with the setting The value is the radius, the number of landslides that occurred in the circular area; ; The variance of the effective stress at all times is ; represents a normal distribution with a mean of 0 and a variance of 1; the encoded result is then input into a single-layer deep learning network to output the latent variable ; is the variance output; It is Hadamard.

[0043] Specifically, the formula first converts the effective stress mean The core idea of ​​this step is to describe the dynamic behavior and uncertainty of the mean output through random mapping. Specifically, the encoding result It follows a normal distribution with a mean given by the distribution function The variance is determined by the logarithmic variance distribution function The advantage of this probabilistic modeling is that it can not only reflect the central trend of the effective stress mean, but also quantify its volatility, especially in areas where environmental conditions change drastically or where slopes have experienced frequent landslides in the past, and can more comprehensively characterize the dynamic characteristics of rock mass units. Mean distribution function The Poisson distribution weight is introduced in the definition of and regional landslide impact radius . Poisson distribution term It is an important tool for describing the spatial and temporal distribution of historical landslide events. determines the spatial scale of landslide events, and It reflects the frequency of landslides in the target slope area. Through this design, the mean distribution function not only describes the current state of the effective stress mean, but also dynamically couples the historical landslide characteristics with the current environmental state. The larger the area, the higher the weight of the Poisson distribution, which indicates that in areas where landslides frequently occur, the cumulative effect of historical disasters has a stronger impact on the current slope stability. Lower or When it is smaller, the influence of Poisson weight is weakened, and the mean distribution function depends more on the current effective stress mean. . Logarithmic variance distribution function This further describes the fluctuation degree of the effective stress mean and its complex relationship with the landslide history. and exponential factor , the variance distribution function explicitly quantifies the regulatory effect of the number of historical landslides and the regional landslide impact radius on the effective stress fluctuation. The higher the value, the larger the dynamic range of the variance, which indicates that the frequent occurrence of historical landslide events will lead to an increase in the uncertainty of the current effective stress. By influencing The magnitude of the landslide reflects the spatial scale of the landslide impact range. The normalization factor of the variance This ensures that effective stress fluctuations at different time points and spatial scales can be reasonably compared, avoiding feature expression biases caused by differences in regional characteristics.

[0044] After defining the mean and variance distribution, the formula further generates potential coding results through random variables. Specifically, the formula introduces Gaussian noise and through The noise and variance distribution are coupled in the form of , generating random samples that conform to the normal distribution. This design is based on the idea of ​​variational autoencoder (VAE), which expresses the encoding result as a joint distribution of mean and volatility. The encoding result generated in this way not only retains the time dynamic trend of effective stress, but also introduces randomness through noise, enhancing the generalization ability of the feature. The final latent variable It is the result of further nonlinear transformation of the encoding result through a single-layer deep learning network. Through this transformation, the latent variables can not only capture the core features of the input data, but also extract hidden patterns that are difficult to directly observe in the deep learning model. The formula combines historical landslide information and current effective stress dynamics through modeling of mean distribution and variance distribution, so that the latent variables can simultaneously reflect the static stability and dynamic evolution trend of the slope. This method can capture the long-term impact of historical events in the slope area. For example, in areas with a high number of landslides, the distribution of latent variables may have greater volatility, thereby predicting a higher risk of instability. At the same time, for areas with relatively stable current environmental conditions, the distribution of latent variables may be more concentrated, reflecting a lower risk of instability. This method of describing dynamic features through probability distribution provides highly robust and sensitive input for subsequent slope stability prediction based on deep learning. Compared with the feature extraction method of the traditional method that only relies on the mean value, the present invention greatly improves the complexity and accuracy of feature expression by jointly modeling the mean and variance distribution of effective stress and introducing historical landslide events and regional environmental factors. The latent variables generated by this dynamic coding method can not only improve the model's prediction accuracy for slope instability risk, but also capture the stability differences between different time points and regions, thus providing a scientific basis for slope monitoring and disaster warning.

[0045] Example 9: In step 3, the latent variables are calculated by the following formula: ; in, Representation calculation The maximum value of Represents the activation function.

[0046] Specifically, the core operation in the formula is to encode the result The maximum value is extracted from the . The stability of expansive rock slopes is the result of the dynamic influence of multiple factors. There is a complex interaction between the stress state of the slope, historical events (such as landslides), and environmental changes (such as rainfall, temperature, etc.). As a key indicator, effective stress will show obvious fluctuations in the time series, and the amplitude of these fluctuations may be closely related to the occurrence of actual instability events. Therefore, the extraction of the maximum value can focus on the important features in the encoded results, that is, those that reflect the most extreme changes experienced by the rock unit or slope area at a specific moment. This maximum value extraction not only improves the model's sensitivity to outliers, but also ensures that the latent variables can capture the most representative state, which is particularly important for predicting the risk of slope instability. Then, the formula normalizes the maximum value and uses The function of normalization is to adjust the range of feature values ​​so that the features of different input data can be compared and processed at the same scale. The normalized value is further transformed by the activation function. It plays a vital role in this process. The activation function compresses or expands the input value into the latent space through nonlinear transformation, ensuring that the latent variables can more accurately express the high-order features in the input data. The stability of the expansive rock slope is affected by many factors, such as changes in environmental conditions, rock mass properties, and external loads. Simple linear models cannot fully capture the complex interactions between these factors. Through the nonlinear transformation of the activation function, the model can automatically learn these complex interactions, thereby achieving higher accuracy and robustness in predicting slope stability.

[0047] The introduction of the activation function not only enhances the model's ability to capture the dynamic behavior of the slope, but also provides stronger expression capabilities for feature learning in subsequent deep learning models. The stability prediction of expansive rock slopes requires not only identifying the stress state of the rock mass in the current environment, but also capturing the potential risk signals in historical data, especially in slope areas where extreme events such as landslides occur. The activation function enables the output of latent variables to be more consistent with these complex, nonlinear relationships, making the model's predictive ability more accurate. For example, an extreme effective stress change at a certain point in time may be a precursor to a landslide, and this change may be highlighted through maximum value extraction during the encoding process, thereby affecting the calculation results of the latent variables. The activation function amplifies this change in a nonlinear way, allowing the model to better reflect the early warning effect of these abnormal data on slope instability. Throughout the process, the latent variables The calculation of is not just a simple transformation of the input features, it is the result of the autoencoder's efficient compression and deep abstraction of the input data. Perform maximum value extraction, normalization and activation function transformation, latent variables It can capture the most important features in the input data and map them into a low-dimensional latent space, so that the deep learning model can use these refined features for subsequent instability probability prediction. The design and calculation of latent variables enable the model to handle the complex environmental dynamics of expansive rock slopes and improve the accuracy and reliability of slope instability risk prediction.

[0048] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that these specific embodiments are only illustrative, and those skilled in the art may omit, replace, and change the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, merging the above method steps so as to perform substantially the same functions in substantially the same manner to achieve substantially the same results is within the scope of the present invention. Therefore, the scope of the present invention is limited only by the appended claims.

Claims

1. A slope stability prediction method based on deep learning, characterized in that: The method comprises: Step 1: Periodically scan the target slope surface with a high-precision laser radar to obtain point cloud data; segment the point cloud data to obtain multiple rock mass units; perform the following operations in the slope area corresponding to each rock mass unit: obtain the geological data of the slope area through geological surveys, and obtain the hydrological and meteorological data of the slope area at multiple continuous times through meteorological station records; Step 2: According to the geological data and hydro-meteorological data of the slope area corresponding to each rock mass unit, the expansion potential energy of the slope area corresponding to each rock mass unit at each time during the water absorption process is calculated; according to the expansion potential energy, the volume strain and the principal strain under isotropic conditions of the slope area corresponding to each rock mass unit at each time are calculated, and according to the volume strain and the principal strain, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated, and then the mean of the effective stress at all times is calculated to obtain the mean output; then the variance of the effective stress at all times is calculated to obtain the variance output; Step 3: Input the mean output as input into the pre-trained deep learning autoencoder, combine it with the variance output, and output the latent variable; calculate the difference between the latent variable and the mean output, and compare the difference with the preset basic threshold to obtain the instability probability of the slope area corresponding to each rock mass unit.

2. The slope stability prediction method based on deep learning according to claim 1, characterized in that: The geological data of the slope area corresponding to each rock unit obtained in step 1 include: rock dry density, rock and soil average particle size, water content, expansion pressure, water density, rock and soil mineral hydration heat, rock elastic modulus, Poisson's ratio, cohesion, internal friction angle, montmorillonite content and quartz content; the rock elastic modulus is measured by collecting rock samples from each rock unit and performing a uniaxial compression test on the rock samples; the cohesion and internal friction angle are measured by collecting rock samples from each rock unit and performing a direct shear test on the rock samples; the Poisson's ratio represents the lateral deformation ratio in another direction when the rock sample is stretched or compressed in one direction.

3. The slope stability prediction method based on deep learning according to claim 2, characterized in that: The hydrological and meteorological data of the slope area corresponding to each rock mass unit obtained in step 1 include: infiltration rate, rainfall and temperature.

4. The slope stability prediction method based on deep learning according to claim 3, characterized in that: In step 1, the surface of the target slope is periodically scanned by a high-precision laser radar to obtain point cloud data; the method of segmenting the point cloud data to obtain multiple rock mass units includes: using the following formula, Adaptive filtering is performed on each point: ； in, For point Point cloud density at ; To search for points within the sphere; is the radius of the search sphere; For point The distance to the nearest neighbor; is the standard deviation of the Gaussian distribution; For point The normal vector at ; is the laser radar line of sight direction vector; then use the following formula to determine the point Neighbors Belongs to a rock mass unit: ； in, is the judgment value, if , then the judgment point Its neighboring points in a rock mass unit; For point Its neighboring points The Euclidean distance of For point Neighbors The normal vector at ; For point Point cloud density at ; is the set angle threshold; is the set distance threshold; is the set density threshold.

5. The slope stability prediction method based on deep learning according to claim 4, characterized in that: In step 2, the expansion potential energy of each rock unit during water absorption at each time is calculated according to the geological data and hydrological and meteorological data of each rock unit by the following formula: ； in, For the The rock mass unit The expansion potential energy during the water absorption process over time; For the The dry density of rock mass in the slope area corresponding to each rock mass unit; For the The average particle size of rock and soil in the slope area corresponding to each rock mass unit; For the The water content of the slope area corresponding to each rock mass unit; For the The expansion pressure of the slope area corresponding to each rock mass unit; For the The bulk density of water in the slope area corresponding to each rock mass unit; For the The hydration heat of rock and soil minerals in the slope area corresponding to each rock mass unit; For the The average height of the slope area corresponding to each rock mass unit; For the The elastic modulus of the rock mass in the slope area corresponding to each rock mass unit; For the Poisson's ratio of the slope area corresponding to each rock mass unit; For the The cohesion of the slope area corresponding to each rock mass unit; For the The internal friction angle of the slope area corresponding to each rock mass unit; For the The montmorillonite content of the slope area corresponding to each rock mass unit; For the Quartz content of the slope area corresponding to each rock mass unit; is the gas constant; is the absolute temperature; for The temperature of time; for rainfall over time; for The penetration rate over time.

6. The slope stability prediction method based on deep learning according to claim 5, characterized in that: In step 2, the volume strain of the slope area corresponding to each rock mass unit at each time and the principal strain under isotropic conditions are calculated according to the expansion potential energy using the following formula: ； in, For the The volume of a rock mass unit; is the volume strain; under the isotropic conditions, the three principal strains are equal, and the calculation formula is: ； in, , and Take the initiative to respond.

7. The slope stability prediction method based on deep learning according to claim 6, characterized in that: In step 2, the effective stress of the slope area corresponding to each rock mass unit at each time is calculated by the following formula: ； in, For the The slope area corresponding to the rock mass unit is Effective stress of time.

8. The slope stability prediction method based on deep learning according to claim 7, characterized in that: In step 3, the mean output is input as the input to the pre-trained deep learning autoencoder. The process of outputting the latent variable specifically includes: assuming that the mean output is ; The mean output is encoded using the following formula: ； in, is the encoding result; is the mean distribution function; is the logarithmic variance distribution function; ;in, is the number of characteristic expansions in the past, and its value is equal to the target slope as the center, with the setting The value is the radius, the number of landslides that occurred in the circular area; ; The variance of the effective stress at all times is ; represents a normal distribution with a mean of 0 and a variance of 1; the encoded result is then input into a single-layer deep learning network to output the latent variable ; is the variance output; It is Hadamard.

9. The slope stability prediction method based on deep learning according to claim 8, characterized in that: In step 3, the latent variables are calculated using the following formula: ； in, Representation calculation The maximum value of Represents the activation function.

Images