Landslide layer structure identification method and system

Abstract

The invention discloses a landslide layer structure identification method and system, landslide three-dimensional diagram information is analyzed by using a target thread, and compared with manual detection, the technical scheme provided by the invention is higher in analysis accuracy. In addition, different target threads are selected for different soil layer structure types for analysis, the analysis accuracy is further improved, and therefore slope structure information can be accurately obtained.

Description

Technical Field

[0001] This disclosure relates to the technical field of landslide layer structure identification, and in particular to a method and system for landslide layer structure identification. Background Technology

[0002] As the core carrier of landslide hazard formation, occurrence, and evolution, the landslide layer structure, its distribution characteristics, lithological combinations, and spatial contact relationships directly determine the stability, sliding mode, and failure intensity of a landslide. It serves as a crucial basis for landslide risk assessment, disaster early warning, and prevention engineering design. Accurate identification of the landslide layer structure (such as soil layer thickness, lithological stratification, interface morphology, and distribution of weak interlayers) provides core data support for analyzing landslide formation mechanisms and predicting sliding risks, significantly enhancing the scientific rigor and targeted approach to landslide hazard prevention and control.

[0003] With the development of geological exploration technology, techniques such as 3D geological modeling, UAV aerial surveying, and ground-penetrating radar detection have been widely applied to landslide layer structure data acquisition, resulting in massive amounts of 3D landslide mapping information. However, current landslide layer structure identification technology still faces many bottlenecks, making it difficult to meet the needs of accurate identification in complex scenarios: First, traditional identification methods rely heavily on manual analysis, which is inefficient and highly subjective. Existing technologies often require professional geologists to manually layer and annotate features based on 3D map information. This not only consumes a lot of manpower and time, but is also easily affected by differences in individual experience and professional level, resulting in poor consistency and large errors in the identification results, making it difficult to meet the needs of large-scale landslide exploration and dynamic monitoring.

[0004] Secondly, there is a lack of targeted hierarchical analysis mechanisms, resulting in insufficient adaptability. Landslide layer structures exhibit significant complexity and diversity. Different soil layer structures (such as large-scale continuous rock layers, small-scale discrete soil layers, and multi-soil layer composite structures) show significant differences in the proportion of feature points, spatial distribution, and positional relationships with adjacent soil layers in 3D image information. Existing recognition technologies mostly employ single, fixed analytical algorithms or models to process various soil layer structures, failing to adapt them differentiated according to the type of soil layer. This leads to insufficient recognition accuracy for complex scenarios such as small-scale soil layers with low feature point proportions and multi-soil layer composite structures, easily resulting in problems such as layer ambiguity, missed judgments, or misjudgments.

[0005] Third, the analysis process lacks a hierarchical design and has weak detail capture capabilities. Identifying landslide layer structures requires considering both the overall spatial distribution and local detailed features. This is especially true in multi-layered composite structures where the target soil layer is strongly correlated with adjacent layers, necessitating multi-stage analysis for accurate localization. Existing technologies often employ a "one-step" analysis process, directly extracting and analyzing features from the overall 3D image. This makes it difficult to effectively distinguish the target soil layer from background information and accurately capture the local details of complex soil layer structures, resulting in identification results that fail to reflect the true boundaries and internal features of the layer structure.

[0006] Fourth, the model training is disconnected from real-world scenarios, resulting in limited generalization performance. Existing recognition models are often trained using single-type example data, failing to fully consider the differences in characteristics across various soil structures. Furthermore, they lack dedicated training and optimization for the parsing sub-threads, leading to insufficient adaptability to new scenarios and soil structure types. Simultaneously, the models lack a dynamic adjustment mechanism corresponding to different soil structure types. When faced with complex and varied 3D landslide images, recognition accuracy tends to drop significantly, making it difficult to meet the needs of practical engineering applications.

[0007] Therefore, how to overcome the technical challenges such as reliance on manual methods, insufficient adaptability, and weak ability to capture details, and how to construct differentiated and hierarchical identification methods based on soil structure types to achieve efficient and accurate analysis of landslide 3D map information has become a core issue that urgently needs to be addressed in the field of landslide structure identification. It is also a key technical support for promoting the transformation of landslide disaster prevention and control from "experience-based judgment" to "precise quantification". Summary of the Invention

[0008] To address the technical problems existing in related technologies, this disclosure provides a method and system for identifying landslide layer structures.

[0009] A method for identifying landslide layer structures includes: Obtain the 3D map information of the landslide to be processed; Based on the slope surface subsurface soil structure information in the landslide 3D map information, a target thread for analyzing the landslide 3D map information is determined, including: determining whether the slope surface subsurface result information is a first target type, obtaining a first analysis result, wherein the first target type is the type in which the proportion of feature points of the target soil structure in the landslide 3D map information is less than a first preset value; and based on the first analysis result, a target thread for analyzing the landslide 3D map information is determined, wherein different soil structure types correspond to different target threads. The target thread is used to analyze the landslide 3D map information; The step of determining the target thread for analyzing the landslide 3D map information based on the first analysis result includes: In response to the slope subsurface result information being a second target type, the target thread is determined to be a first target thread corresponding to the second target type, where the second target type is a type in which the proportion of feature points of the target soil structure in the landslide 3D map information is not less than a first preset value. Alternatively, in response to the slope subsurface result information being a first target type, it is determined whether the first target type is multiple soil layer structure types, and a second analysis result is obtained, wherein the multiple soil layer structure types are adjacent soil layer structures that have a relatively fixed position relative to the target soil layer structure. The step of determining the target thread for analyzing the landslide 3D map information based on the second analysis result includes: in response to the first target type being a soil structure type, determining the target thread as a second target thread corresponding to the soil structure type, wherein the soil structure type is a soil structure that does not have an adjacent soil structure whose position is relatively fixed with the target soil structure; and in response to the first target type being multiple soil structure types, determining the target thread as a third target thread corresponding to the multiple soil structure types.

[0010] In one independently implemented embodiment, the slope subsurface result information is related to the size of the target soil layer structure and / or the positional relationship between the target soil layer structure and adjacent soil layer structures. Before determining the target thread for analyzing the landslide 3D map information based on the slope subsurface soil layer structure information in the landslide 3D map information, the method further includes: Receive training tasks on soil structure types; In response to the slope subsurface result information training task, soil structure types are set for the target soil structure corresponding to the landslide 3D map information; Alternatively, determine the similarity between the landslide 3D map information and the historical geological structure information in the dataset, wherein the historical geological structure information is information that has been analyzed, and the dataset is annotated with the slope subsurface soil structure information in each of the historical geological structure information; The slope subsurface soil structure information in the target historical geological structure information that meets the similarity requirements is identified as the soil structure type corresponding to the target soil structure in the landslide 3D map information.

[0011] In one independently implemented embodiment, the first target thread includes a first target analysis sub-thread and a first analysis sub-thread, and the second target thread includes a second target analysis sub-thread and a second analysis sub-thread. The step of analyzing the landslide 3D map information using the target threads includes: The landslide 3D map information is visually analyzed using the first or second visual analysis sub-thread to obtain several visual analysis ranges. The analysis results of the landslide 3D map information are obtained by using either the first analysis sub-thread or the second target analysis sub-thread to analyze the target analysis range.

[0012] In one independently implemented embodiment, the third target thread includes a third target analysis sub-thread, a fourth target analysis sub-thread, and a third analysis sub-thread. The analysis of the landslide 3D map information using the target thread includes: The third visual analysis sub-thread is used to perform visual analysis on the landslide 3D map information to obtain several undetermined detection ranges covering the adjacent soil layer structure; The fourth visual resolution sub-thread is used to perform visual resolution on the undetermined detection range to obtain the visual resolution range covering the target soil structure; The third analysis sub-thread is used to analyze the target measurement range to obtain analysis results about the landslide three-dimensional map information.

[0013] In one independently implemented embodiment, the target thread includes a visual analysis sub-thread and an analysis sub-thread. The visual analysis sub-thread is used to perform visual analysis on the landslide 3D map information to obtain visual analysis results. The analysis sub-thread is used to analyze the landslide 3D map information based on the visual analysis results to obtain analysis results of the landslide 3D map information. The method includes a training process for each target thread, the training process including: For each type of slope subsurface result information, first example information covering the target soil structure corresponding to the slope subsurface result information is obtained, and the target analysis sub-thread corresponding to the slope subsurface result information is trained using the first example information; and second example information is obtained, and different sub-soil structures of the target soil structure in the second example information are labeled to obtain labeling information. The information segmentation thread is trained using the aforementioned annotation information; The sub-thread used for feature extraction in the information division thread is identified as the feature extraction sub-thread of the analysis sub-thread. The analysis sub-thread is trained using third-paradigm information, which includes the slope result data to be analyzed. The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects.

[0014] A method and system for identifying landslide layer structures are disclosed. By analyzing three-dimensional landslide data using target threads, the accuracy of the analysis achieved is higher than that of manual inspection. Furthermore, selecting different target threads for different soil layer structures further improves the accuracy of the analysis, thereby enabling the accurate acquisition of slope structure information.

[0015] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit this disclosure. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the principles of this application.

[0017] Figure 1 A flowchart illustrating a landslide layer structure identification method provided in this application embodiment; Detailed Implementation

[0018] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0019] Based on the above, please refer to the following: Figure 1 This is a flowchart illustrating a landslide layer structure identification method provided in an embodiment of this application. Further, a landslide layer structure identification method may specifically include the content described in steps S11-S13.

[0020] Step S11: Obtain the three-dimensional map information of the landslide to be processed.

[0021] Step S12: Based on the information on the underground soil structure of the slope surface in the landslide 3D map, determine the target thread for analyzing the landslide 3D map information. Different soil structure types correspond to different target threads.

[0022] In some disclosed embodiments, the information on the underground soil structure of the slope can be determined based on the size of the soil structure in which the slope result data is located, or based on the characteristics of the soil structure in which the slope result data is located, or simultaneously based on both the size of the soil structure in which the slope result data is located and the characteristics of the soil structure in which the slope result data is located.

[0023] Among them, different target threads corresponding to different soil structure types can be either the same soil structure for the target threads but different thread parameters, or different soil structure for the target threads.

[0024] In some disclosed embodiments, the target thread corresponding to the landslide 3D map information can be one or multiple different target threads. For example, if there are two target threads corresponding to the landslide 3D map information, the landslide 3D map information is input into these two target threads for analysis, resulting in two analysis results. The analysis of the landslide 3D map information by each target thread can be executed in parallel.

[0025] Step S13: Analyze the landslide 3D map information using the target thread.

[0026] Specifically, the landslide 3D map information is used as input to the target thread, which processes the landslide 3D map information to obtain the analysis results.

[0027] The above-described solution, by using a target thread to analyze the three-dimensional landslide image information, achieves higher accuracy compared to manual detection. Furthermore, selecting different target threads for analysis of different soil structure types further enhances the accuracy of the analysis, thereby enabling the accurate acquisition of slope structure information.

[0028] In some disclosed embodiments, the soil structure type is related to the size of the target soil structure and / or the positional relationship between the target soil structure and adjacent soil structures. In some application scenarios, the soil structure type is related to the size of the target soil structure. In some application scenarios, the soil structure type is related to the positional relationship between the target soil structure and adjacent soil structures. In some application scenarios, the soil structure type is related to both the size of the target soil structure and the positional relationship between the target soil structure and adjacent soil structures. Adjacent soil structures refer to soil structures whose position is relatively fixed relative to the target soil structure. Before executing step S12, the following steps may also be performed: The system receives a training task on soil structure types. In response to this task, it sets the soil structure type for the target soil structure corresponding to the landslide 3D map information. The target soil structure is the soil structure located within the slope surface result data to be analyzed in the landslide 3D map information.

[0029] In some application scenarios, the target soil structure is categorized into large and small target soil structure types based on its size. The small target soil structure type is defined as a soil structure whose proportion of feature points in the landslide 3D map is less than a pre-defined proportion. The large target soil structure type is defined as a soil structure whose proportion of feature points in the landslide 3D map is not less than a pre-defined proportion. In other application scenarios, the target soil structure is categorized into multiple soil structure types and a single soil structure type based on its positional relationship with adjacent soil structures. Multiple soil structure types can include adjacent soil structures with relatively fixed positions relative to the target soil structure, such as the U-shaped pin example mentioned above. A single soil structure type can be characterized by the absence of adjacent soil structures with relatively fixed positions relative to the target soil structure, such as a loose nut defect. Since nuts are widely used in the entire contact network, it is difficult to summarize the fixed adjacent soil structures with relatively fixed positions relative to the target soil structure.

[0030] Soil structure types are assigned to target soil structures based on their size, type, and positional relationship with adjacent soil structures. For example, these can be categorized into large target soil structure types, single-soil-structure small target types, and multiple soil-structure small target types.

[0031] In some disclosed embodiments, the following steps may also be performed before performing step S12: The similarity between the landslide 3D map information and the historical geological structure information in the dataset is determined. The historical geological structure information refers to information that has already been analyzed, and the dataset annotates the soil structure types of the target soil layer in each historical geological structure information where the slope result data to be analyzed is located. The dataset can be set in the execution device. Then, the slope subsurface soil structure information in the target historical geological structure information that meets the similarity requirements is determined as the soil structure type corresponding to the target soil structure in the landslide 3D map information. In some other embodiments, if there is no target historical geological structure information that meets the similarity requirements with the landslide 3D map information, a prompt is issued so that the user can issue a soil structure type training task based on the prompt. Then, the execution device responds to the soil structure type training task and sets the soil structure type for the target soil structure corresponding to the landslide 3D map information. The specific method of setting the soil structure type for the target soil structure corresponding to the landslide 3D map information in response to the soil structure type training task can be found in the previous embodiment, and will not be repeated here.

[0032] By identifying the soil structure type based on the soil layer structure of the slope result data to be analyzed, a target thread for analyzing the three-dimensional landslide map information is determined. This allows for the reference to the soil layer structure information of the slope result data to be analyzed, thereby improving the accuracy of the analysis of the three-dimensional landslide map information.

[0033] An embodiment of the analysis method of this application illustrates a partial flowchart of step S12. Step S12 may specifically include the following steps: Step S121: Determine whether the soil structure type is the first target type and obtain the first analysis result.

[0034] Step S122: Based on the first analysis results, determine the target thread for analyzing the landslide 3D map information.

[0035] By determining the corresponding target thread based on the proportion of feature points of the target soil structure in the 3D landslide map, the analysis can be performed by selecting the appropriate target thread based on the size of the target soil structure, thereby improving the accuracy of the analysis.

[0036] In some disclosed embodiments, the method for determining the target thread for analyzing the landslide 3D map information based on the first analysis result may be as follows: In response to the soil structure type being the second target type, the target thread is determined to be the first target thread corresponding to the second target type. The second target type is defined as the type in which the proportion of feature points of the target soil structure in the landslide 3D map information is not less than a first pre-set value. Specifically, the second target type can be the aforementioned large target type.

[0037] In response to the soil structure type being the first target type, it is determined whether the first target type comprises multiple soil structure types, resulting in a second analysis result. Multiple soil structure types refer to adjacent soil structures that are relatively fixed in position to the target soil structure. Specifically, a soil structure that simultaneously belongs to both the first target type and multiple soil structure types is considered a sub-target type of multiple soil structures; a soil structure that belongs to both the first target type and a single soil structure type is considered a single sub-target type.

[0038] Then, based on the results of the second analysis, the target thread for analyzing the three-dimensional landslide map information was determined.

[0039] By selecting the corresponding target thread based on whether the target soil structure has multiple soil structure types, the positional relationship between the target soil structure and adjacent soil structures can be considered, thereby improving the accuracy of the analysis.

[0040] The method for determining the target thread for analyzing the landslide 3D map information based on the second analysis results may include: In response to the first target type being a soil structure type, a second target thread corresponding to that soil structure type is determined. Here, a soil structure type is defined as a soil structure that does not have any adjacent soil structures with a relatively fixed position to the target soil structure. That is, when the target soil structure is the aforementioned sub-target type of soil structure, the target thread for analyzing the landslide 3D map information is determined as the second target thread.

[0041] In response to the first target type being multiple soil structure types, the target thread is determined to be the third target thread corresponding to the multiple soil structure types.

[0042] That is, when the target soil structure is one of the above-mentioned multiple soil structure sub-targets, the target thread for analyzing the landslide 3D map information is determined as the third target thread.

[0043] By selecting the corresponding target thread based on whether the target soil structure has multiple soil structure types, the positional relationship between the target soil structure and adjacent soil structures can be considered, thereby improving the accuracy of the analysis.

[0044] The first target thread includes a first target parsing and testing sub-thread and a first analysis sub-thread. The second target thread includes a second target parsing and testing sub-thread and a second analysis sub-thread. The third target thread includes a third target parsing and testing sub-thread, a fourth target parsing and testing sub-thread, and a third analysis sub-thread. The thread parameters in each target parsing and testing sub-thread can be the same or different, and the thread parameters in each analysis sub-thread can also be the same or different.

[0045] In some disclosed embodiments, step S13 above may include the following steps: In this embodiment, after determining that the target thread for analyzing the landslide 3D map information includes a first target thread, a first visual analysis sub-thread is used to perform visual analysis on the landslide 3D map information, obtaining several visual analysis ranges. Specifically, the "several" in this embodiment can be one or more; for example, there can be one or two visual analysis ranges. Specifically, a visual analysis range refers to the information range covering the target soil layer structure. That is, multiple information ranges covering the target soil layer structure can be obtained through visual analysis. The first analysis sub-thread analyzes the visual analysis ranges to obtain the analysis results of the landslide 3D map information. Specifically, several visual analysis ranges are input into the first analysis sub-thread, which analyzes each visual analysis range to obtain the analysis results for each range. These analysis results are then concatenated to obtain the analysis results of the landslide 3D map information.

[0046] In this process, after determining that the target thread for analyzing the landslide 3D image information includes a second target thread, the second visual analysis sub-thread performs visual analysis on the landslide 3D image information, obtaining several visual analysis ranges. Specifically, the visual analysis range refers to the information range covering the target soil layer structure. That is, multiple information ranges covering the target soil layer structure can be obtained through visual analysis. The second analysis sub-thread analyzes the visual analysis ranges to obtain the analysis results of the landslide 3D image information. That is, several visual analysis ranges are input into the second analysis sub-thread, which analyzes each visual analysis range to obtain the analysis results of each visual analysis range. The analysis results of each visual analysis range are then stitched together to obtain the analysis results of the landslide 3D image information. The second visual analysis sub-thread in the second target thread has a stronger detection capability for target soil layer structures of smaller target types.

[0047] By first visually analyzing soil layers with large structures or a single type of soil structure to obtain several visual analysis ranges, and then analyzing these ranges, the two-level analysis method can further improve the accuracy of the analysis.

[0048] After determining that the target thread used to analyze the landslide 3D map information includes a third target thread, step S13 above may include the following steps: A third visual analysis sub-thread performs visual analysis on the landslide 3D map information, obtaining several undetermined detection ranges covering adjacent soil structures. Here, adjacent soil structures refer to those with relatively fixed positions relative to the target soil structure. After determining the positions of each adjacent soil structure through visual analysis of the landslide 3D map information, the third visual analysis sub-thread determines the size of the undetermined detection range based on the positional relationship between each adjacent soil structure and the target soil structure, as well as the size of each adjacent soil structure and the target soil structure. This ensures that the undetermined detection range covers both the adjacent soil structures and the target soil structure, facilitating visual analysis of the undetermined detection range by a fourth visual analysis sub-thread.

[0049] Then, the fourth visual analysis sub-thread performs visual analysis on the target detection range to obtain the visual analysis range covering the target soil structure. Specifically, each target detection range is determined as the input of the fourth visual analysis sub-thread. After performing visual analysis on each target detection range, the fourth visual analysis sub-thread obtains the position of each target soil structure within the target detection range. Based on the position of each target soil structure, the target detection range is trimmed to obtain the corresponding visual analysis range.

[0050] Finally, the third analysis sub-thread analyzes the target measurement range to obtain the analysis results of the landslide 3D map information. Specifically, each target measurement range is defined as the input to the third analysis sub-thread, which analyzes each target measurement range to obtain the analysis results for each range. These analysis results are then concatenated to obtain the analysis results of the landslide 3D map information.

[0051] By first determining the undetermined detection range for multiple soil structure types, then determining the target detection range from the undetermined detection range, and finally analyzing the target detection range, this three-level analysis method can further improve the accuracy of the analysis.

[0052] The first, third, and fourth analytical sub-threads can be the same analytical sub-thread capable of locating multiple target soil structures. That is, the analytical sub-thread is trained using training examples covering multiple soil structures, enabling it to detect various soil structures.

[0053] As mentioned above, the target thread includes a visual analysis sub-thread and an analysis sub-thread. The visual analysis sub-thread performs visual analysis on the landslide 3D map information to obtain visual analysis results. The visual analysis results include several visual analysis ranges. The analysis sub-thread analyzes the landslide 3D map information based on the visual analysis results to obtain analysis results for the landslide 3D map information. That is, the analysis sub-thread analyzes several visual analysis ranges to obtain analysis results for the landslide 3D map information.

[0054] In some disclosed embodiments, the analysis method provided in this disclosure further includes a training process for each target thread. The training process may include the following steps: For each soil structure type, first example information covering the target soil structure corresponding to that soil structure type is obtained, and the visual analysis sub-thread corresponding to that soil structure type is trained using the first example information. As mentioned above, the target soil structure is the soil structure where the slope result data to be analyzed is located. The target soil structure covered in the first example information may or may not contain the slope result data to be analyzed. For training the visual analysis sub-thread, the first example information needs to label the target soil structure, but does not need to label the slope result data to be analyzed. As mentioned above, the visual analysis sub-thread can be a thread that performs visual analysis on the target soil structure, or it can be a thread that performs visual analysis on adjacent soil structures. Therefore, example information covering adjacent soil structures can be used to train the visual analysis sub-thread used to detect adjacent soil structures. Each soil structure type can correspond to one or more target soil structures. That is, different target soil structures belonging to the same soil structure type can be used to train the visual analysis sub-thread.

[0055] The sub-thread used for feature extraction in the information partitioning thread is designated as the feature extraction sub-thread of the analysis sub-thread. That is, the sub-thread used for feature extraction in the information partitioning thread is designated as the pre-training thread for the analysis sub-thread, and the analysis sub-thread is trained accordingly.

[0056] By using examples of slope result data to be analyzed, the target thread corresponding to each slope result data to be analyzed is trained, so that the trained target thread can analyze the corresponding slope result data.

[0057] The information partitioning thread is trained first, and then a sub-thread for feature extraction is formed within the information partitioning thread.

[0058] The feature extraction sub-thread is designated as the analysis sub-thread, enabling it to consider the soil structural characteristics of the slope surface data being analyzed. It should be understood that this application is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for identifying landslide layer structure, characterized in that, include: Obtain the 3D map information of the landslide to be processed; Based on the slope surface subsurface soil structure information in the landslide 3D map information, a target thread for analyzing the landslide 3D map information is determined, including: determining whether the slope surface subsurface result information is a first target type, obtaining a first analysis result, wherein the first target type is the type in which the proportion of feature points of the target soil structure in the landslide 3D map information is less than a first preset value; and based on the first analysis result, a target thread for analyzing the landslide 3D map information is determined, wherein different soil structure types correspond to different target threads. The target thread is used to analyze the landslide 3D map information; The step of determining the target thread for analyzing the landslide 3D map information based on the first analysis result includes: In response to the slope subsurface result information being a second target type, the target thread is determined to be a first target thread corresponding to the second target type, where the second target type is a type in which the proportion of feature points of the target soil structure in the landslide 3D map information is not less than a first preset value. Alternatively, in response to the slope subsurface result information being a first target type, it is determined whether the first target type is multiple soil layer structure types, and a second analysis result is obtained, wherein the multiple soil layer structure types are adjacent soil layer structures that have a relatively fixed position relative to the target soil layer structure. The step of determining the target thread for analyzing the landslide 3D map information based on the second analysis result includes: in response to the first target type being a soil structure type, determining the target thread as a second target thread corresponding to the soil structure type, wherein the soil structure type is a soil structure that does not have an adjacent soil structure whose position is relatively fixed with the target soil structure; and in response to the first target type being multiple soil structure types, determining the target thread as a third target thread corresponding to the multiple soil structure types.

2. The method according to claim 1, characterized in that, The slope subsurface result information is related to the size of the target soil layer structure and / or the positional relationship between the target soil layer structure and adjacent soil layer structures. Before determining the target thread for analyzing the landslide 3D map information based on the slope subsurface soil layer structure information in the landslide 3D map information, the method further includes: Receive training tasks on soil structure types; In response to the slope subsurface result information training task, soil structure types are set for the target soil structure corresponding to the landslide 3D map information; Alternatively, determine the similarity between the landslide 3D map information and the historical geological structure information in the dataset, wherein the historical geological structure information is information that has been analyzed, and the dataset is annotated with the slope subsurface soil structure information in each of the historical geological structure information; The slope subsurface soil structure information in the target historical geological structure information that meets the similarity requirements is identified as the soil structure type corresponding to the target soil structure in the landslide 3D map information.

3. The method according to claim 1, characterized in that, The first target thread includes a first target analysis sub-thread and a first analysis sub-thread, and the second target thread includes a second target analysis sub-thread and a second analysis sub-thread. The step of analyzing the landslide 3D map information using the target threads includes: The landslide 3D map information is visually analyzed using the first or second visual analysis sub-thread to obtain several visual analysis ranges. The analysis results of the landslide 3D map information are obtained by using either the first analysis sub-thread or the second target analysis sub-thread to analyze the target analysis range.

4. The method according to claim 1, characterized in that, The third target thread includes a third target analysis sub-thread, a fourth target analysis sub-thread, and a third analysis sub-thread. The analysis of the landslide 3D map information using the target thread includes: The third visual analysis sub-thread is used to perform visual analysis on the landslide 3D map information to obtain several undetermined detection ranges covering the adjacent soil layer structure; The fourth visual resolution sub-thread is used to perform visual resolution on the undetermined detection range to obtain the visual resolution range covering the target soil structure; The third analysis sub-thread is used to analyze the target measurement range to obtain analysis results about the landslide three-dimensional map information.

5. The method according to any one of claims 1-4, characterized in that, The target thread includes a visual analysis sub-thread and an analysis sub-thread. The visual analysis sub-thread is used to perform visual analysis on the landslide 3D map information to obtain visual analysis results. The analysis sub-thread is used to analyze the landslide 3D map information based on the visual analysis results to obtain analysis results of the landslide 3D map information. The method includes a training process for each target thread, and the training process includes: For each type of slope subsurface result information, first example information covering the target soil structure corresponding to the slope subsurface result information is obtained, and the target analysis sub-thread corresponding to the slope subsurface result information is trained using the first example information; and second example information is obtained, and different sub-soil structures of the target soil structure in the second example information are labeled to obtain labeling information. The information segmentation thread is trained using the aforementioned annotation information; The sub-thread used for feature extraction in the information division thread is identified as the feature extraction sub-thread of the analysis sub-thread. The analysis sub-thread is trained using third-paradigm information, which includes the slope result data to be analyzed.

6. A landslide layer structure identification system, characterized in that, It includes a processor and a memory that communicate with each other, the processor being used to read a computer program from the memory and execute it to implement the method of any one of claims 1-5.

Images