Small and medium-sized high-position collapse geological disaster hidden danger identification method and device based on microtopographic features and storage medium

Abstract

The invention provides a micro-topographic feature-based small and medium-sized high-position collapse geological disaster hidden danger identification method and device and a storage medium, and relates to the technical field of geological disaster hidden danger identification. The recognition method comprises the following steps: acquiring a disaster-prone micro-topographic feature index system of a slope unit to be recognized in a working area; the index system comprises a plurality of first-level indexes including a slope unit form, and each first-level index comprises a plurality of second-level indexes; inputting a multi-dimensional geometric feature vector formed on the basis of each secondary index after standardization processing into the trained hidden danger recognition model, outputting a reconstruction feature vector with the same input dimension by the hidden danger recognition model, and calculating a reconstruction error by comparing the difference between the input and the output; and obtaining a potential collapse index based on the reconstruction error and the weighted fusion of each secondary index, wherein the larger the value is, the higher the possibility of the collapse disaster of the slope is. By adopting the identification method provided by the invention, small and medium-sized high-position collapse hidden dangers in a high-susceptibility geological disaster area can be efficiently and accurately distinguished.

Description

Technical Field

[0001] This application relates to the field of geological hazard identification technology, and in particular to a method, device, and storage medium for identifying small and medium-sized high-altitude landslide geological hazard hazards based on micro-topographic features. Background Technology

[0002] Small and medium-sized high-altitude landslides are typical geological hazards in my country, mostly occurring in the upper and middle parts of steep slopes. They are characterized by high concealment, suddenness, random spatial distribution, and large impact and destructive force, making hazard identification difficult and a key focus and challenge in geological hazard prevention and control. The characteristics of their landslide source areas being far from the slope toe and having high potential energy make it difficult to identify this type of landslide hazard using traditional methods, seriously threatening the safety of life and property.

[0003] Small and medium-sized high-altitude landslides exhibit several unique characteristics: First, they are highly concealed. Landslide source areas are usually located on high, steep cliffs that are difficult to observe directly from human activity. Vegetation cover or terrain obstruction further increases the difficulty of identification by the naked eye and traditional survey methods. Second, it is characterized by its suddenness; the process from rock mass loosening to final instability often takes only a few minutes to a few hours, and conventional monitoring methods are difficult to effectively capture the precursors. Third, the energy is concentrated and the impact range is large. Although the volume of the landslide is generally between several hundred and tens of thousands of cubic meters, which is relatively small, the high potential energy is converted into huge kinetic energy, often forming impact-debris flow, which causes devastating damage to linear projects such as houses and roads in the middle and lower parts.

[0004] Currently, developing large-scale, high-precision, and intelligent hazard identification technologies is one of the key, challenging, and globally critical issues in this field. The development of observation technologies such as UAV aerial surveying and lidar has enabled the rapid acquisition of high-precision regional DEM data, providing solid support for the extraction and quantification of slope micro-topographic features, which are the most critical controlling factors for the occurrence of such disasters.

[0005] However, existing identification methods mostly focus on macroscopic indicators, neglecting the indicative role of micro-topography. They also suffer from shortcomings such as incomplete feature selection, unreasonable weight allocation, and the inability to perform only binary judgments, thus failing to accurately quantify the degree of hazard. Therefore, there is an urgent need for an efficient and feasible identification method that can systematically extract micro-topographic features, accurately quantify hazard levels, and overcome the shortcomings of existing technologies to meet the needs of precise geological disaster prevention and control. Summary of the Invention

[0006] In view of this, the present application provides a method, device and storage medium for identifying small and medium-sized high-altitude landslide geological hazard hazards based on micro-topographic features, so as to at least solve one of the problems in the prior art.

[0007] In a first aspect, embodiments of this application provide a method for identifying potential small-to-medium-sized high-altitude landslide geological hazards based on micro-topographic features, the identification method comprising: Obtain the disaster-prone micro-topographic feature index system of the slope unit to be identified in the work area. The disaster-prone micro-topographic feature index system includes multiple primary indicators, including slope unit morphology, boot-shaped slope morphology, slope shoulder turning zone, slope surface concave cavity, slope vertical free face and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators. The multidimensional geometric feature vectors formed by the standardized secondary indicators are input into the trained hazard identification model. The hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and the output. The potential collapse index of the slope unit to be identified is obtained by weighted fusion of the reconstruction error and various secondary indicators; the larger the value of the potential collapse index, the higher the probability of a collapse disaster on the slope.

[0008] Secondly, this application also provides a device for identifying small-to-medium-sized high-altitude landslide geological hazard hazards based on micro-topographic features, the device comprising: Memory is used to store executable instructions for a computer; A processor, used to implement the identification method of the above-mentioned technical solution when executing computer-executable instructions stored in the memory.

[0009] Thirdly, embodiments of this application also provide a storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the identification method of the above-described technical solution.

[0010] Based on the micro-topographic feature-based method for identifying small-to-medium-sized high-altitude landslide geological hazard risks proposed in this application, a risk-prone micro-topographic feature index system for the slope units to be identified within the work area is obtained. This index system includes six primary indicators, such as slope unit morphology and boot-shaped slope morphology, as well as all secondary indicators corresponding to each primary indicator, constructing a complete micro-geomorphic feature data foundation. The extracted secondary indicators are standardized to eliminate dimensional differences and construct multi-dimensional geometric feature vectors, which are then input into a pre-trained hazard identification model. By comparing the reconstructed feature vector output by the model with the input vector, the reconstruction error is accurately calculated, quantifying the degree of feature anomaly. A weighted fusion mechanism for each secondary indicator is introduced, and the potential landslide index of the slope unit to be identified is calculated based on the reconstruction error. The degree of hazard is judged according to the index magnitude. This scheme can solve the problem of identifying hidden high-altitude landslide hazards, utilize high-precision DEM data to quickly identify hazards, provide precise support for geological disaster prevention and control, improve the scientific nature of prevention and control, and reduce disaster losses.

[0011] Additional advantages, objectives, and features of this application will be set forth in part in the description which follows, and will in part become apparent to those skilled in the art upon review of the following description, or may be learned by practice of the application. The objectives and other advantages of this application can be realized and obtained by means of the structures specifically pointed out in the specification and drawings.

[0012] Those skilled in the art will understand that the purposes and advantages that can be achieved with this application are not limited to those specifically described above, and that the above and other purposes that this application can achieve will be more clearly understood from the following detailed description. Attached Figure Description

[0013] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, do not constitute a limitation thereof. The components in the drawings are not drawn to scale but are merely for illustrating the principles of this application. For ease of illustration and description of certain parts of this application, corresponding portions in the drawings may be enlarged, i.e., may appear larger relative to other components in an exemplary device actually manufactured according to this application. In the drawings: Figure 1 This is a flowchart of a method for identifying potential small-to-medium-sized high-altitude landslide geological hazards based on micro-topographic features, according to an embodiment of this application. Figure 2 This is a schematic diagram of a small-to-medium-sized high-altitude landslide hazard zone in a method for identifying small-to-medium-sized high-altitude landslide geological hazard hazards based on micro-topographic features according to an embodiment of this application; Figure 3 This is a schematic diagram of the slope plane projection features in a method for identifying small and medium-sized high-altitude landslide geological hazards based on micro-topographic features according to an embodiment of this application; Figure 4 This is a schematic diagram of the slope unit morphology index in a method for identifying small and medium-sized high-altitude landslide geological hazards based on micro-topographic features according to an embodiment of this application; Figure 5 This is a schematic diagram of a boot-shaped slope in a method for identifying potential geological hazards of small and medium-sized high-altitude landslides based on micro-topographic features, according to an embodiment of this application. Figure 6 This is a schematic diagram of the shoulder turning zone and its characteristics in a method for identifying small and medium-sized high-altitude landslide geological hazards based on micro-topographic features according to an embodiment of this application. Figure 7 This is a schematic diagram of the cavity and vertical free-face index of the slope unit in the method for identifying small and medium-sized high-altitude landslide geological hazards based on micro-topographic features according to an embodiment of this application. Figure 8This is a schematic diagram of a small-to-medium-sized high-altitude landslide geological hazard identification device based on micro-topographic features according to an embodiment of this application; Figure 9 This is a schematic diagram of a small-to-medium-sized high-altitude landslide geological hazard identification system based on micro-topographic features, according to an embodiment of this application. Detailed Implementation

[0014] The purposes and functions of this application, as well as the methods for achieving these purposes and functions, will be clarified by referring to exemplary embodiments. However, this application is not limited to the exemplary embodiments disclosed below; it can be implemented in various forms. The specification is merely intended to help those skilled in the art to comprehensively understand the specific details of this application.

[0015] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or combinations thereof.

[0016] Ordinal numbers such as “first” and “second” used in this application are merely identifiers and have no other meaning, such as a specific order. Moreover, for example, the term “first component” does not imply the existence of a “second component”, and the term “second component” does not imply the existence of a “first component”.

[0017] It should be noted that the terms “up,” “down,” “front,” “back,” “left,” “right,” “inner,” “outer,” and similar expressions used in this article are for illustrative purposes only and are not intended to be limiting.

[0018] In the following description, embodiments of the present application will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0019] First, refer to Figure 1 This application describes a method 100 for identifying potential small-to-medium-sized high-altitude landslide geological hazards based on micro-topographic features, according to an embodiment of this application. For example... Figure 1 As shown, the identification method 100 may include steps S110 to S130, as detailed below: In step S110, the disaster-prone micro-topographic feature index system of the slope unit to be identified in the working area is obtained. The disaster-prone micro-topographic feature index system includes multiple primary indicators, including slope unit morphology, boot-shaped slope morphology, slope shoulder turning zone, slope surface concave cavity, slope vertical free face and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators.

[0020] In step S120, the multidimensional geometric feature vector formed based on the standardized secondary indicators is input into the trained hazard identification model. The hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and the output.

[0021] In step S130, the potential collapse index of the slope unit to be identified is obtained by weighted fusion of the reconstruction error and various secondary indicators; the larger the value of the potential collapse index, the higher the probability of a collapse disaster occurring on the slope.

[0022] As described above, this identification method can quickly identify and quantify the severity of small and medium-sized high-altitude landslide hazards, providing accurate data support and decision-making basis for geological disaster prevention and control. By comprehensively capturing the characteristics of slope unit hazards through a systematic micro-geomorphological index system, and combining model reconstruction errors with weighted fusion, the level of landslide hazards can be accurately quantified, greatly improving the efficiency of hazard investigation and the scientific nature of prevention and control, and effectively reducing casualties and property losses caused by disasters.

[0023] For small to medium-sized high-altitude landslides with strong instantaneous disaster characteristics, static identification factor (topography) analysis can be used for calculation. Compared with dynamic identification factors (deformation magnitude, crack expansion), static identification factor analysis is more practical, and the data collection of static identification factors has higher reliability and data quality. In particular, based on high-precision topographic data, the micro-topographic features of slopes prone to high-altitude landslides can be finely characterized and identified, thereby effectively capturing the potential locations of high-altitude landslide hazards.

[0024] The topographic constraints on small and medium-sized landslides at high altitudes mainly include the slope of the boot-shaped terrain with a steep upper part and a gentle lower part. The two specific micro-geomorphic features that indicate the potential landslide source area, namely the "concave cavity on the slope surface" and the "vertical free face of the slope" that delineates the unloading boundary, are key micro-geomorphic disaster-prone topographic features.

[0025] Slope surface cavities refer to localized negative topographic features developed on steep slopes, appearing as inward-concave "cavities" or "niches." They are an important micro-topographic precursor to high-altitude landslides. The morphological characteristics of slope surface cavities are a distinct inward-concave arc-shaped curve along the two-dimensional profile of the slope. In three-dimensional space, they appear as a concave area bounded on three sides and open on one side. The formation of slope surface cavities is typically caused by geological forces such as differential weathering, rock unloading, erosion of weak interlayers, or previous small-scale collapses. Cavity areas usually exhibit a specific combination of positive profile curvature (convex) and negative plane curvature (concave), with a negative openness value.

[0026] A vertical free face on a slope refers to a nearly vertical or reverse-dipping steep face on a slope that provides space for landslides and collapses. It is a prerequisite for landslides, characterized by extremely steep local slopes, typically greater than 60°, and even approaching or exceeding 90° (vertical or inverted). This face leaves the overlying or rearward rock and soil completely exposed, losing lateral support. A vertical free face on a slope can be quantitatively extracted based on high-precision DEM data using two parameters: slope and aspect. In a two-dimensional profile, it appears as a steep sloping edge with a rapidly increasing slope value; in three-dimensional space, by calculating the rate of change of aspect, linear bands where the aspect changes sharply can be identified. These bands often correspond to the boundaries of the unstable rock mass or the free face itself.

[0027] By conducting detailed identification, extraction, and comprehensive judgment of topographic features such as boot-shaped terrain with steep upper slopes and gentle lower slopes, surface concave cavities of slopes, and free-faced slopes, and further utilizing hazard identification models, we can accurately identify small and medium-sized high-altitude landslide hazards.

[0028] First, the acquired high-precision point cloud data from airborne LiDAR and other sources undergoes preprocessing calculations, including point cloud denoising, filtering, and classification, to generate a high-resolution digital elevation model (DEM). Subsequently, terrain feature enhancement calculations are performed, deriving a series of basic terrain parameter raster layers for micro-topography identification based on the DEM. Core parameters include regional terrain geometric indices such as slope / aspect, terrain curvature, and terrain openness. If a high-precision DEM is unavailable, 3D point cloud data can be generated using UAV oblique photogrammetry, and slope units can be extracted through point cloud segmentation.

[0029] Furthermore, a quantitative disaster-prone terrain identification index called the "Potential Landslide Index" and its quantitative calculation model were proposed. This index can automatically calculate the degree of fit between slope morphology and typical boot-shaped terrain, thereby accurately locating disaster-prone areas. Ultimately, this improves the identification of small to medium-sized high-altitude landslide hazards with strong concealment from "experience-based judgment" to "data-driven, model-based identification," enhancing the scientific rigor, accuracy, and efficiency of the identification process.

[0030] Furthermore, a disaster-prone micro-topographic feature index system was constructed, which covers multiple dimensions of indicators, including slope unit morphology, boot-shaped topographic morphology, slope shoulder transition zone, slope surface cavities and vertical free face, as well as micro-topographic combination morphology. The indicators are closely related to the disaster-causing mechanical mechanism of high-level landslides, which is characterized by "steep upper slope and gentle lower slope" and "unloading at free face." For example, "cavities" indicate unloading and tension zones, and "free face" indicates instability boundaries.

[0031] During the feature extraction process, the basic shape of the slope unit can be obtained through high-precision DEM to obtain the boot-shaped index, thereby determining whether it is a boot-shaped terrain. For boot-shaped terrain, UAV point cloud data is further collected to construct a three-dimensional model, and the slope cavity and vertical free face are identified to obtain micro-geomorphological indicators and their combination relationships.

[0032] The following will describe in detail the contents of each step of the identification method according to the embodiments of this application.

[0033] In the embodiments of this application, reference is made to Figure 2 , Figures 3 to 7 As shown, in step S110, a disaster-prone micro-topographic feature index system for the slope units to be identified within the work area is obtained. This system includes multiple primary indicators, such as slope unit morphology, boot-shaped slope morphology, slope shoulder transition zone, slope surface cavities, vertical exposed slope surfaces, and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators. Specifically, in step S110, each primary indicator includes multiple secondary indicators, specifically: Slope unit index feature extraction: like Figure 4 As shown, secondary indices of slope unit morphology can be calculated and extracted based on the three-dimensional geometric characteristics of the slope unit profile. These secondary indices can include the main profile slope length, main profile slope width, length-to-width ratio of the slope unit along the main profile direction, total elevation difference from the top to the bottom of the landslide, length-to-height ratio, area under the profile curve, area under the profile straight line, and area ratio.

[0034] The length and width of the slope can be extracted by taking the main profile direction of the slope as the length direction and its perpendicular direction as the width direction, and the length-to-width ratio can be calculated.

[0035] Along the main profile of the slope, interpolation can be performed based on the accuracy of the DEM to generate a curve showing the change in elevation along the main profile. The slope height is then extracted based on this curve, and the length-to-height ratio is calculated simultaneously.

[0036] By extracting the area below the main profile line, we can connect the highest and lowest points of the slope to obtain a straight line segment, extract the area below this straight line segment, and then calculate the area ratio.

[0037] Boot-shaped terrain feature extraction: like Figure 5 As shown, slope inflection points can be identified based on the three-dimensional geometric features of the slope profile, dividing the slope into steep sill and gentle slope zones. Secondary indices for the boot-shaped slope morphology are calculated and extracted. These secondary indices may include boot-shaped terrain determination values, steep sill length, steep sill height, gentle slope length, gentle slope height, steep sill height ratio, steep sill slope gradient, and gentle slope gradient.

[0038] If the ratio of the area under the profile curve to the area under the straight line connecting the beginning and end points of the profile curve is less than a specified threshold (e.g., 0.8), then the area can be determined to be boot-shaped terrain.

[0039] By identifying the inflection point of the curvature of the main profile, the length, height, angle, and area under the profile line of the steep slope area and the gentle slope area can be extracted, and the proportional relationship between the above indicators can be calculated.

[0040] The inflection point can be obtained in the following ways: Calculate the curvature Ki at each point along the main profile; Traverse all points on the profile and select the point with the minimum local curvature (satisfying...) > < ), which are the concave points on the cross-section; If only one concave point is obtained, that point is the turning point between the steep slope area and the gentle slope area; if multiple concave points are obtained, the concave point with the longest perpendicular distance from the line connecting the start and end points of the main profile is selected as the turning point.

[0041] Feature extraction of the shoulder transition zone: like Figure 6 As shown, the secondary indicators of the shoulder transition zone can include the curvature of the transition point, the width of the transition zone, and the elevation difference of the transition zone. The maximum indices can be identified by calculating the sum of the profile and planar curvatures, thus obtaining the curvature of the shoulder transition zone. Strip-shaped areas with a change rate > 20% can be extracted based on the slope aspect rate raster, and the width of the shoulder transition zone can be calculated. The elevation difference can be extracted along the shoulder line, and the elevation difference of the shoulder transition zone can be calculated.

[0042] Cavity feature extraction: like Figure 7 As shown, a DEM model generated from 3D point cloud data can be used to extract downward-facing triangular faces. Cavities can then be obtained through region growing. The secondary indices of the extracted slope surface cavities can be calculated, including the secondary indices of the cavities themselves. These secondary indices can include the number of cavities, cavity depth, cavity area, lower edge height of the cavity, cavity profile curvature, and cavity planar curvature.

[0043] The cavity range can be determined using a region growing method: Any vertex with its normal pointing downwards is taken as a seed, added to a queue, and marked as visited. The neighborhood can be grown outwards following a breadth-first search principle. If the angle between the normal vector of an adjacent point and the seed point is less than a threshold, and the adjacent point is also a concave point, then the adjacent point is expanded and added to the queue. If none of the adjacent vertices meet the above conditions, or if the mesh boundary is reached, then expansion can stop. If the cavity range obtained through growth is greater than a certain threshold, it can be determined as a valid cavity.

[0044] The cavity properties can be extracted using the following steps: If multiple cavities exist, the one with the largest depth can be selected, and its attribute value recorded. Obtain the edge points of the cavity and construct the least squares fitting plane; Cavity depth acquisition: Calculate the maximum distance from all Mesh vertices to the plane, and use it as the cavity depth; Obtaining the area of ​​the concave cavity: Calculate the area of ​​the polygonal edge of the concave cavity projected onto the above plane; Lower edge height acquisition: Calculate the minimum Z value of the edge point; Curvature acquisition: Calculate the average planar curvature and average cross-sectional curvature of all vertices within the concave cavity.

[0045] Feature extraction of vertical free surface indicators: like Figure 7 As shown, a DEM model generated from 3D point cloud data can be used to obtain the vertical free face through connectivity analysis by specifying the normal vector direction and the included angle threshold, and then calculating the secondary indices of the extracted vertical free face of the slope. The secondary indices of the vertical free face of the slope can include the number of vertical free faces, the slope of the free face, the height of the free face, the area of ​​the free face, the height of the upper edge of the free face, and the rate of change of slope aspect.

[0046] An area with a continuous slope greater than (a specified slope threshold) and an area greater than (a specified area threshold) can be considered a valid free surface.

[0047] Vertical free faces can be obtained through connectivity analysis: calculate the normal vector of each triangular face. Triangular faces with slopes that meet certain conditions can be filtered by using a threshold angle between the normal vector and the horizontal plane. Connectivity analysis can then be performed on these satisfying triangular faces to obtain one or more connected regions. The area of ​​each connected region and the average curvature of all vertices can be calculated. Connected regions with an area greater than a specified threshold and an absolute value of the average curvature less than a specified threshold can be considered as vertical free faces.

[0048] If there are multiple free faces, the one with the largest area can be selected and its attribute value recorded.

[0049] Micro-topographic feature extraction: Secondary indicators for micro-topographic morphology include the cavity-free surface distance and the combined area ratio. The cavity-free surface distance can be calculated by determining the shortest Euclidean distance between the lower edge of the cavity and the boundary of the free surface based on spatial overlay analysis. The combined area ratio of the micro-topographic morphology can be calculated by extracting the total area of ​​the cavity and free surface within the steep slope area and comparing it to the area of ​​the steep slope area.

[0050] If both a concave cavity and a vertical surface are present, the distance from the lower end of the concave cavity to the upper end of the vertical surface can be calculated. The areas of both and their proportion of the steep slope area can then be calculated.

[0051] The specific primary and secondary indicators of the disaster-prone micro-topography characteristic index system are shown in Table 1. Table 1

[0052] In step S120, the standardization process specifically refers to: Each secondary indicator can be standardized to ensure model stability.

[0053] in, As a secondary indicator, and Let be the mean and variance of the i-th indicator, respectively.

[0054] Furthermore, the hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and the output, specifically referring to: A hazard identification model can be constructed using autoencoders from deep learning. The core structure of this model can consist of two parts: an encoder f and a decoder g. The encoder compresses the input x into a latent representation z, and the decoder reconstructs the input... ; In the formula, W1 and W2 are weight matrices, b1 and b2 are bias terms, and σ is the ReLU activation function, which realizes complex feature transformation. Calculate the reconstruction error. In the formula, E represents the reconstruction error. The larger E is, the more abnormal the input features are.

[0055] Furthermore, the identification method may also include training a hazard identification model to obtain a well-trained hazard identification model. The process of training the hazard identification model may include: To obtain samples of the vulnerable micro-topographic feature index system for small and medium-sized high-altitude landslide hazard slope units within the work area, the sample set of multidimensional geometric feature vectors formed based on all vulnerable micro-topographic feature index system samples can be used as the training set, which is then input into the initial autoencoder model. It is necessary to standardize the multidimensional geometric feature vectors formed by all vulnerable micro-topographic feature index system samples to eliminate the influence of dimensions.

[0056] The encoder and decoder of the initial autoencoder model transform each geometric feature vector in the training set into a reconstructed feature vector of the same dimension. The initial autoencoder model can employ a deep autoencoder network structure built with fully connected layers. The encoder progressively reduces the dimension to a 5-dimensional latent space, while the decoder symmetrically recovers the original dimension.

[0057] The initial autoencoder model can be trained unsupervised, using the Adam optimizer to minimize the reconstruction error. When the reconstruction error stops decreasing or begins to increase over multiple consecutive training epochs, or when the model reaches the preset maximum number of training epochs, the autoencoder model can be considered a well-trained hazard identification model. Dropout layers and L2 regularization can be added during training to prevent overfitting, and early stopping can be used to optimize training efficiency.

[0058] Model training can be stopped early using early stopping and the maximum number of training epochs. Specifically, during training, the model monitors the reconstruction error (mean squared error, MSE, can be selected) on an independent validation set. If the validation set error stops decreasing or even starts to increase over several consecutive epochs, training is stopped to prevent overfitting.

[0059] Additionally, to prevent excessively long training times, a maximum number of training epochs can be preset (e.g., Epochs=100), after which the training can be forcibly terminated. The ultimate goal is to find the weights that best perform on the validation set, ensuring generalization ability.

[0060] Specifically, in step S130, the potential collapse index of the slope unit to be identified is obtained by weighted fusion based on the reconstruction error and by introducing various secondary indices. Specifically, it refers to: The reconstruction error can be converted into an exponential value, while a weighted fusion of secondary indicators is introduced to enhance interpretability. The weight W of each quantitative indicator is defined. i The weights can be determined by the statistical significance of historical samples (such as logistic regression) and / or expert experience; In the formula, I is the combination of the weighted sum and the reconstruction error. and To harmonic the parameters, the contribution of the control weighted sum and the error term is adjusted through training; The identification method may further include: normalizing the potential landslide index. In the formula, The normalized index of the potential landslide index, I, can be normalized to [0,1]. The closer the index is to 1, the higher the risk of high-risk landslides.

[0061] Furthermore, a higher potential landslide index indicates a greater likelihood of landslides occurring on the slope. Specifically, it refers to: Risk levels can be classified based on the relationship between the potential collapse index and a set threshold: a potential collapse index ≤ a low-risk threshold indicates a high-risk level; a potential collapse index between low and high thresholds indicates a medium-risk level; and a potential collapse index > a high-risk threshold indicates a low-risk level.

[0062] The discrimination thresholds may include the area ratio threshold, the cavity range threshold, the normal vector angle threshold, the area threshold of the vertical free face, the average curvature threshold of the vertical free face, the high-risk threshold of the potential landslide index, and the low / high reconstruction error threshold. All discrimination thresholds can be calibrated and determined based on the actual geological conditions and historical landslide case data within the working area.

[0063] Furthermore, by combining the results of field investigations, data on false alarms and missed alarms in slope units can be collected as supplementary training samples to feed back into and optimize the algorithm parameters, index weights, and discrimination thresholds of the hazard identification model, thereby improving the accuracy of hazard identification.

[0064] refer to Figure 8 This application also provides an identification device 200 for implementing the identification method 100 according to the embodiments of this application. The identification device 200 includes a processor 210 and a memory 220. The identification device 200 may include one or more processors 210 and one or more memories 220. The memory 220 stores an executable program that is run by the processor 210. When the executable program is run by the processor 210, it causes the processor 210 to execute the identification method 100 described above according to the embodiments of this application.

[0065] The processor 210 may be a central processing unit (CPU) or other processing units with data processing capabilities and / or instruction execution capabilities.

[0066] The memory 220 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 210 may execute the program instructions to implement the client functions (implemented by the processor) in the embodiments of this application described herein, and / or other desired functions. Various applications and various data may also be stored in the computer-readable storage medium, such as various data used and / or generated by the applications.

[0067] The identification device 200 may also include input and output devices, which are interconnected via a bus system and / or other forms of connection mechanisms. It should be noted that... Figure 8 The components and structure of the identification device 200 shown are merely exemplary and not limiting; the identification device 200 may also have other components and structures as needed.

[0068] The input device can be a device used by a user to input commands, and can include one or more of a keyboard, mouse, microphone, and touchscreen. Furthermore, the input device can also be any interface for receiving information.

[0069] The output device can output various information (e.g., images or sounds) to the outside (e.g., a user), and may include one or more of a display, speaker, etc. Furthermore, the output device can also be any other device with output functionality.

[0070] For example, the identification device 200 for implementing the identification method 100 according to the embodiments of this application can be applied to terminal devices (such as mobile phones), tablet computers, laptop computers, ultra-mobile personal computers (UMPCs), handheld computers, netbooks, personal digital assistants (PDAs), wearable devices (such as smartwatches, smart glasses, or smart helmets), augmented reality (AR) devices, virtual reality (VR) devices, smart home devices, in-vehicle computers, and other electronic devices. The embodiments of this application do not impose any limitations on this.

[0071] Those skilled in the art can understand the specific operation of the identification device 200 of the identification method 100 according to the embodiments of this application in conjunction with the content described above. For the sake of brevity, the specific details will not be repeated here, but only some main operations of the processor 210 will be described.

[0072] In one embodiment of this application, when the executable program is run by the processor 210, the processor 210 performs the following steps: acquiring a disaster-prone micro-topographic feature index system for the slope unit to be identified within the working area. The disaster-prone micro-topographic feature index system includes multiple primary indicators, such as slope unit morphology, boot-shaped slope morphology, slope shoulder transition zone, slope surface concave cavity, slope vertical free face, and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators. The multidimensional geometric feature vector formed based on the standardized secondary indicators is input into a trained hazard identification model. The hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and output. Based on the reconstruction error and by introducing a weighted fusion of the secondary indicators, a potential landslide index for the slope unit to be identified is obtained. The larger the potential landslide index value, the higher the probability of a landslide disaster occurring on the slope.

[0073] The above exemplarily illustrates an identification method 100 according to an embodiment of this application. The following, in conjunction with... Figure 9 The following describes an identification system 300 provided in another aspect of an embodiment of this application.

[0074] refer to Figure 9 This describes an identification system 300 used to implement embodiments of the present application. The identification system 300 may include a first acquisition module 310, a first adjustment module 320, and a construction module 330. Wherein: The first acquisition module 310 is used to: acquire the disaster-prone micro-topographic feature index system of the slope unit to be identified in the work area. The disaster-prone micro-topographic feature index system includes multiple primary indicators, including slope unit morphology, boot-shaped slope morphology, slope shoulder turning zone, slope surface concave cavity, slope vertical free face and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators. The first adjustment module 310 is used to: input the multi-dimensional geometric feature vector formed based on the standardized secondary indicators into the trained hazard identification model; the hazard identification model outputs a reconstructed feature vector with the same dimension as the input; and calculate the reconstruction error by comparing the difference between the input and the output. The construction module 330 is used to: obtain the potential collapse index of the slope unit to be identified by weighted fusion based on the reconstruction error and by introducing various secondary indicators; the larger the value of the potential collapse index, the higher the probability of a collapse disaster occurring on the slope.

[0075] Furthermore, according to embodiments of this application, this application also provides a storage medium on which a computer program is stored. When the computer program is run by a processor, it is used to execute corresponding steps of the identification method 100 of this application. The storage medium may, for example, include a memory card of a smartphone, a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0076] Furthermore, according to embodiments of this application, this application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the identification method 100 of embodiments of this application.

[0077] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of this application. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of this application. All such changes and modifications are intended to be included within the scope of this application as claimed in the appended claims.

[0078] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0079] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0080] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. For example, in the claims, any one of the claimed embodiments can be used in any combination.

[0081] It should be noted that the above embodiments are illustrative of this application and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0082] The above description is merely a specific embodiment or illustration of the embodiments of this application. The scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. The scope of protection of this application shall be determined by the scope of the claims.

Claims

1. A method for identifying potential small-to-medium-sized high-altitude landslide geological hazards based on micro-topographic features, characterized in that, The identification method includes: Obtain the disaster-prone micro-topographic feature index system of the slope unit to be identified in the work area. The disaster-prone micro-topographic feature index system includes multiple primary indicators, including slope unit morphology, boot-shaped slope morphology, slope shoulder turning zone, slope surface concave cavity, slope vertical free face and micro-topographic combination morphology. Each primary indicator includes multiple secondary indicators. The multidimensional geometric feature vectors formed by the standardized secondary indicators are input into the trained hazard identification model. The hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and the output. The potential collapse index of the slope unit to be identified is obtained by weighted fusion of the reconstruction error and various secondary indicators; the larger the value of the potential collapse index, the higher the probability of a collapse disaster on the slope.

2. The identification method according to claim 1, characterized in that, Each primary indicator includes multiple secondary indicators, specifically: The secondary indicators of slope unit morphology are calculated and extracted based on the three-dimensional geometric characteristics of the slope unit profile line. These include the main profile slope length, main profile slope width, length-to-width ratio of the slope unit along the main profile direction, total height difference from the top to the bottom of the landslide, length-to-height ratio, area under the profile curve, area under the profile straight line, and area ratio. Based on the three-dimensional geometric features of the slope profile, the slope inflection point is identified, and the slope is divided into steep sill area and gentle slope area. The secondary indicators of boot-shaped slope morphology are calculated and extracted, including boot-shaped terrain judgment value, length of steep sill area, height of steep sill area, length of gentle slope area, height of gentle slope area, height ratio of steep sill area, slope of steep sill area and slope of gentle slope area. The curvature of the turning zone of the shoulder is obtained by calculating the sum of the curvature of the profile and the plane to identify the maximum point; the strip area with a change rate > 20% is extracted based on the slope aspect change rate raster, and the width of the turning zone of the shoulder is calculated; the elevation difference is extracted along the shoulder line, and the elevation difference of the turning zone of the shoulder is calculated. Based on the DEM model generated from 3D point cloud data, the triangular facets with downward normal are extracted, and concave cavities are obtained through region growing. The secondary indices of the extracted concave cavities on the slope surface are calculated, including the number of concave cavities, the depth of concave cavities, the area of ​​concave cavities, the height of the lower edge of concave cavities, the curvature of the concave cavity profile, and the curvature of the concave cavity plane. Based on the DEM model generated from 3D point cloud data, the normal vector direction and included angle threshold are specified. The vertical free face is obtained through connectivity analysis. The secondary indicators of the vertical free face of the slope are calculated and extracted, including the number of vertical free faces, the slope of the free face, the height of the free face, the area of ​​the free face, the height of the upper edge of the free face, and the rate of change of slope aspect. Based on spatial overlay analysis, the shortest Euclidean distance between the lower edge of the cavity and the boundary of the free surface is calculated to obtain the cavity-free surface distance of the micro-topographic combination morphology; the total area of ​​the cavity and free surface in the steep slope area is extracted, and the ratio of the area of ​​the steep slope area to the combined area ratio of the micro-topographic combination morphology is calculated to obtain the combined area ratio of the micro-topographic combination morphology.

3. The identification method according to claim 2, characterized in that, The identification of slope inflection points specifically refers to: Calculate the curvature at each point on the profile line. ; Traverse all points to obtain the local minimum curvature value, i.e., the concave point on the profile; if only one concave point is obtained, it is the turning point between the steep slope area and the gentle slope area; if there are multiple concave points, select the one with the longest perpendicular distance from the starting point to the ending point; and / or The cavity range is determined using a region growing method: Any vertex with a downward normal is taken as a seed, added to a queue, and marked as visited; the neighborhood is grown outwards according to a breadth-first search principle; if the angle between the normal vector of an adjacent point and the seed point is less than a threshold, and the adjacent point is also a concave point, then the adjacent point is expanded and added to the queue; if all adjacent vertices do not meet the above conditions, or the mesh boundary is reached, then expansion stops; if the cavity range obtained by growth is greater than a certain threshold, it is determined to be a valid cavity; and / or The process of obtaining a vertical free-standing face through connectivity analysis specifically involves: calculating the normal vector of each triangular face; filtering by a threshold angle between the normal vector and the horizontal plane to obtain triangular faces with slopes that meet the conditions; performing connectivity analysis on the triangular faces that meet the conditions to obtain one or more connected regions; calculating the area of ​​the connected regions and the average curvature of all vertices; and selecting connected regions whose area is greater than a specified threshold and whose absolute value of the average curvature is less than a specified threshold as vertical free-standing faces.

4. The identification method according to claim 1, characterized in that, Standardization processing specifically refers to: Standardize each secondary indicator to ensure model stability. In the formula, As a secondary indicator, Let be the mean of the i-th quantitative indicator. Let be the standard deviation of the i-th quantitative indicator.

5. The identification method according to claim 1, characterized in that, The hazard identification model outputs a reconstructed feature vector with the same dimension as the input. The reconstruction error is calculated by comparing the difference between the input and the output. Specifically, this means: A hazard identification model is constructed using an autoencoder from deep learning. The core structure of the hazard identification model consists of two parts: an encoder f and a decoder g. The encoder compresses the input x into a latent representation z, and the decoder reconstructs the input. ; In the formula, W1 and W2 are weight matrices, b1 and b2 are bias terms, and σ is the ReLU activation function, which realizes complex feature transformation. Calculate the reconstruction error. In the formula, E represents the reconstruction error. The larger E is, the more abnormal the input features are.

6. The identification method according to claim 1, characterized in that, The identification method further includes training a hazard identification model to obtain a trained hazard identification model; wherein, the process of training the hazard identification model includes: The sample of the disaster-prone micro-topographic feature index system of small and medium-sized high-altitude landslide disaster slope units in the work area is obtained. The sample set of multi-dimensional geometric feature vectors formed based on all disaster-prone micro-topographic feature index system samples is used as the training set and the training set is input into the initial autoencoder model. The encoder and decoder of the initial autoencoder model transform each geometric feature vector in the training set into a reconstructed feature vector of the same dimension; The initial autoencoder model is trained in an unsupervised manner, and the Adam optimizer is used to minimize the reconstruction error. When the reconstruction error no longer decreases or begins to increase in consecutive rounds, or when the training model reaches the preset maximum number of training rounds, the autoencoder model is determined to be a well-trained hazard identification model.

7. The identification method according to claim 1, characterized in that, The potential collapse index of the slope unit to be identified is obtained by weighting and fusing various secondary indicators based on the reconstruction error. Specifically, it refers to: Define the weight W for each quantitative indicator. i The weights are determined through the statistical significance of historical samples and / or expert experience; In the formula, I is the combination of the weighted sum and the reconstruction error. and To harmonic the parameters, the contribution of the control weighted sum and the error term is adjusted through training; The identification method further includes: normalizing the potential landslide index. In the formula, The normalized index of the potential landslide index, I, can be normalized to [0,1]. The closer the index is to 1, the higher the risk of high-risk landslides.

8. The identification method according to claim 7, characterized in that, The identification method also includes classifying the potential landslide index into three risk levels: high, medium, and low. Specifically: Risk levels are classified according to the relationship between the potential collapse index and the set threshold: a potential collapse index ≤ a low-risk threshold is a high-risk level, a potential collapse index between low and high thresholds is a medium-risk level, and a potential collapse index > a high-risk threshold is a low-risk level. The discrimination thresholds include the area ratio threshold, the cavity range threshold, the normal vector angle threshold, the area threshold of the vertical free surface, the average curvature threshold of the vertical free surface, the high-risk threshold of the potential landslide index, and the low / high threshold of the reconstruction error; all discrimination thresholds are calibrated and determined based on the actual geological conditions and historical landslide case data in the working area.

9. A small-to-medium-sized high-altitude landslide geological hazard identification device based on micro-topographic features, characterized in that, The identification device includes: Memory is used to store executable instructions for a computer; A processor, when executing computer-executable instructions stored in the memory, implements the identification method according to any one of claims 1 to 8.

10. A storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the identification method according to any one of claims 1 to 8.

Images