A landslide susceptibility prediction method, device and equipment

Abstract

The embodiment of the present application provides a landslide proneness prediction method, device and equipment, which are applied to the computer technology field, and the method comprises the following steps: obtaining basic data of a slope region, wherein the basic data is related to the topography, hydrology and human engineering of the slope region; the ground surface of the slope region is divided to form a plurality of boundary-closed slope units; based on the basic data, a multi-dimensional characteristic value of each slope unit is determined, wherein the multi-dimensional characteristic value is related to the topography, hydrology and human engineering of the slope unit; based on the positional relationship between the slope unit and a landslide region where a landslide has occurred, a plurality of slope units containing the multi-dimensional characteristic value are prepared to form a training sample; an initial prediction model is trained in combination with the training sample to obtain a target prediction model, wherein the target prediction model is used to predict the probability of a landslide occurring in an input slope unit; and the target prediction model is called to predict the slope unit which has not occurred a landslide.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a method, apparatus and equipment for predicting landslide susceptibility. Background Technology

[0002] Landslide susceptibility mapping (LSM) is the core of the geological disaster risk management system. Currently, the industry has shifted from traditional qualitative heuristic models and binary statistical models to a data-driven machine learning paradigm dominated by support vector machines (SVM), random forests (RF), and deep learning (such as CNN).

[0003] When constructing datasets for such evaluation models, the evaluation process is essentially a typical "positive-unlabeled learning" problem. Since landslide cataloging typically only includes confirmed landslide points (positive samples), existing technologies generally employ global random sampling based on regular grid cells, low-slope static thresholding, or Euclidean distance-based buffering methods to forcibly delineate negative sample regions when acquiring non-landslide data (negative samples).

[0004] Despite continuous improvements in the fitting ability of algorithmic models, simply pursuing higher algorithmic complexity has reached a bottleneck. Existing technologies suffer from the following significant shortcomings in terms of input data quality and model interpretability: (1) Negative sample screening is blind and easily introduces "label noise": Traditional global random sampling ignores the lag and concealment of landslide development; while the low slope threshold method often fails in complex geological environments. For example, in the Loess Plateau, the flat loess plateau is prone to concealed erosion and collapse due to the development of vertical joints, strong water sensitivity, and disturbance from human activities such as slope cutting and building. The forced hard threshold is very likely to misclassify potentially unstable slopes in high-risk areas as negative samples, causing the model to learn incorrect feature patterns and produce serious prediction bias.

[0005] (2) Traditional sampling strategies are topologically incompatible with "slope units": Most existing negative sample screening strategies rely on the geometric uniformity of regular grid units. When the evaluation carrier is upgraded to a "slope unit" with high geometric heterogeneity and complex topological relationships, traditional Euclidean distance buffering or pixel threshold sampling methods are difficult to apply directly. At present, there is still a lack of intelligent negative sample screening methods specifically designed for the heterogeneity of attributes and spatial autocorrelation within slope units.

[0006] (3) "Black box" models lack the ability to explain individual mechanisms: As the complexity of machine learning increases, the problem of its lack of interpretability becomes increasingly prominent. Traditional variable importance ranking can only provide global patterns and cannot provide local explanations for each prediction sample. Geological decision-makers cannot know for sure whether the probability of disaster in a specific geomorphic location (such as the edge of a plateau and the bottom of a gully) is dominated by micro-topography, hydrological conditions or human activities, making it difficult to reveal deep-seated differences in disaster-causing patterns. Summary of the Invention

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a landslide susceptibility prediction method, comprising: Obtain basic data for the slope area, which is related to the topography, hydrology, and human engineering of the slope area; The surface of the slope region is divided into multiple closed-bound slope units; Based on the aforementioned basic data, multidimensional feature values ​​are determined for each slope unit. These multidimensional feature values ​​are related to the topography, hydrology, and human engineering of the slope unit. Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are prepared to form training samples. The initial prediction model is trained using the training samples to obtain the target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit. The target prediction model is invoked to predict the slope units that have not yet experienced a landslide.

[0008] In one embodiment, the step of preparing training samples from multiple slope units containing the multidimensional feature values ​​based on the positional relationship between the slope unit and the landslide area where a landslide has occurred includes: Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are divided into positive samples and a first unlabeled sample. The positive sample indicates that the slope area is located within the landslide area, and the first unlabeled sample indicates that the slope area is located outside the landslide area. The training samples are formed by processing the positive samples and the first unlabeled samples using a spying method.

[0009] In one embodiment, the step of using a spying method to process the positive samples and the first unlabeled samples to form the training samples includes: A portion of the positive samples is extracted as spy samples, and the spy samples are mixed into the first unlabeled samples to form a second unlabeled sample; The training samples are formed based on the positive samples and a second unlabeled sample mixed with the spy samples.

[0010] In one embodiment, the step of calling the target prediction model to predict the slope units that have not yet experienced a landslide includes: The target prediction model is invoked to predict the spy sample and the first unlabeled sample; The method further includes: Determine the predicted probability of the spy sample output by the target prediction model; The confidence threshold is determined based on the predicted probability of the spy sample; The predicted probability of each of the first unlabeled samples is compared based on the confidence threshold, and the non-landslide area is determined based on the comparison result.

[0011] In one embodiment, the predicted probability of each of the first unlabeled samples is compared based on the confidence threshold, and non-slide areas are determined based on the comparison results, including: The slope unit corresponding to the first unlabeled sample whose predicted probability is lower than the confidence threshold is determined as a non-slope unit; All the aforementioned non-landslide units constitute the non-landslide area.

[0012] In one embodiment, obtaining the basic data of the slope region includes: Obtain basic topographic data of the slope area, including slope, aspect, relative elevation difference, elevation, and landform type; Obtain hydrological distribution maps, including normalized water index; Obtain data on human engineering activities, including normalized difference vegetation index, normalized difference building index, land use type, and distance from roads.

[0013] In one embodiment, dividing the surface of the slope region into multiple closed-boundary slope units includes: Determine the DEM data of the slope area; The DEM data of the slope region is processed based on the terrain spatial analysis algorithm to adaptively generate multiple slope units with closed boundaries.

[0014] In one embodiment, determining the multidimensional feature values ​​of each slope unit based on the basic data includes: Determine the basic data corresponding to each slope unit; Spatial overlay analysis and feature extraction are performed on the basic data of the slope units to obtain multidimensional feature values ​​for each slope unit. The multidimensional feature values ​​include slope, aspect, relative elevation difference, elevation, landform type, normalized vegetation index, normalized water index, normalized building index, land use type, and distance from the road.

[0015] Another embodiment of the present invention also provides a landslide susceptibility prediction device, comprising: The module is used to obtain basic data of the slope area, which is related to the topography, hydrology, and human engineering of the slope area. The division module is used to divide the ground surface of the slope area into multiple slope units with closed boundaries; The determination module is used to determine the multidimensional feature values ​​of each slope unit based on the basic data, wherein the multidimensional feature values ​​are related to the topography, hydrology, and human engineering of the slope unit; The preparation module is used to prepare multiple slope units containing the multidimensional feature values ​​to form training samples based on the positional relationship between the slope unit and the landslide area where a landslide has occurred. The training module is used to train an initial prediction model by combining the training samples to obtain a target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit. The prediction module is used to call the target prediction model to predict the slope units that have not yet experienced a landslide.

[0016] Another embodiment of the present invention also provides an electronic device, comprising: One or more processors; Memory, configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the landslide susceptibility prediction method as described above.

[0017] Other features and advantages of this application will be set forth in the following description. The objectives and other advantages of this application can be realized and obtained through the structures particularly pointed out in the written description and drawings.

[0018] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the landslide susceptibility prediction method in an embodiment of the present invention.

[0021] Figure 2This is a flowchart illustrating a landslide susceptibility prediction method according to another embodiment of the present invention.

[0022] Figure 3 This is a flowchart illustrating a landslide susceptibility prediction method in another embodiment of the present invention.

[0023] Figure 4 This is a structural block diagram of the landslide susceptibility prediction device in an embodiment of the present invention. Detailed Implementation

[0024] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but these are not intended to limit the scope of the invention.

[0025] It should be understood that various modifications can be made to the embodiments disclosed herein. Therefore, the following description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope of this disclosure will be apparent to those skilled in the art.

[0026] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present disclosure and, together with the general description of the disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the disclosure.

[0027] These and other features of the invention will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0028] It should also be understood that although the invention has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of the invention, which have the features described in the claims and are therefore all within the scope of protection defined herein.

[0029] The above and other aspects, features and advantages of this disclosure will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0030] Specific embodiments of the present disclosure are described thereafter with reference to the accompanying drawings; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure and can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the present disclosure. Therefore, the specific structural and functional details disclosed herein are not intended to be limiting, but merely to serve as the basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of substantially any suitable detailed structures.

[0031] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in still another embodiment,” all of which may refer to one or more of the same or different embodiments according to this disclosure.

[0032] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] like Figure 1 As shown, this embodiment of the invention provides a landslide susceptibility prediction method, including: S1: Obtain basic data of the slope area, which is related to the topography, hydrology, and human engineering of the slope area; S2: Divide the surface of the slope area into multiple closed-bound slope units; S3: Determine the multidimensional feature value of each slope unit based on the basic data. The multidimensional feature value is related to the topography, hydrology, and human engineering of the slope unit. S4: Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are prepared to form training samples; S5: Train an initial prediction model using the training samples to obtain a target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit; S6: Invoke the target prediction model to predict the slope units that have not yet experienced a landslide.

[0034] This embodiment addresses the technical shortcomings of existing landslide susceptibility assessment technologies, such as "blind selection of negative samples leading to label noise," "incompatibility between traditional sampling methods and slope units," and "lack of geological interpretability of black-box models." It proposes the above-mentioned technical solution, which breaks through the existing technical bias of "simply pursuing algorithmic complexity" and shifts to a new path of "data quality optimization and mechanism interpretation." By introducing a semi-supervised learning paradigm, it collaboratively utilizes a small number of labeled landslide slope units and a massive number of unlabeled slope units to intelligently mine the potential distribution structure of the data. This allows for the extraction of "high-purity" negative samples that fit the topological structure of the slope units without relying on subjective human thresholds.

[0035] In one embodiment, obtaining the basic data of the slope region includes: S101: Obtain basic topographic data of the slope area, including slope, aspect, relative elevation difference, elevation, and landform type; S102: Obtain hydrological distribution maps, including the normalized water index; S103: Obtain data on human engineering activities, including normalized difference vegetation index, normalized difference building index, land use type, and distance from roads.

[0036] Based on the above data, the system can construct a basic database for the slope area. Simultaneously, in this embodiment, the system can also combine the obtained geological survey records of the slope area to establish a landslide database for that slope area, which contains the location information of known landslides in the slope area.

[0037] Furthermore, such as Figure 2 As shown, dividing the surface of the slope region into multiple closed-bound slope units includes: S201: Determine the DEM data of the slope area; S202: The DEM data of the slope area is processed based on the terrain spatial analysis algorithm to adaptively generate multiple slope units with closed boundaries.

[0038] In this embodiment, the obtained DEM data of the slope area is used to adaptively divide the continuous land surface of the slope area into multiple closed-bound slope units using a topographic spatial analysis algorithm. Each slope unit conforms to the actual topographic undulations, so using it as the basic unit for subsequent landslide susceptibility prediction can improve prediction accuracy.

[0039] Furthermore, determining the multidimensional feature values ​​of each slope unit based on the basic data includes: S301: Determine the basic data corresponding to each slope unit; S302: Perform spatial overlay analysis and feature extraction on the basic data of the slope unit to obtain multidimensional feature values ​​for each slope unit. The multidimensional feature values ​​include slope, aspect, relative elevation difference, elevation, landform type, normalized vegetation index, normalized water index, normalized building index, land use type, and distance from the road.

[0040] For example, but not limited to, the spatial zonal statistics function of a geographic information system (ArcGIS software) can be used to perform spatial overlay analysis and feature extraction on the basic data of each slope unit, using the slope unit as the evaluation unit. This allows for the quantitative calculation and extraction of various environmental factors for each slope unit, i.e., the multidimensional feature values. These environmental factors, as described above, specifically include: slope, aspect, relative elevation difference, elevation, landform type, normalized difference vegetation index (NDVI), normalized difference water index (NDWI), normalized difference building index (NDBI), land use type, and distance from roads. Other values ​​may also be included; the specific values ​​are not unique.

[0041] In another embodiment, the step of preparing training samples from multiple slope units containing the multidimensional feature values ​​based on the positional relationship between the slope unit and the landslide area where a landslide has occurred includes: S401: Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are divided into positive samples and a first unlabeled sample. The positive sample indicates that the slope area is located within the landslide area, and the first unlabeled sample indicates that the slope area is located outside the landslide area. S402: The positive samples and the first unlabeled samples are processed using a spying method to form the training samples.

[0042] The solution in this embodiment, within the framework of a defined slope unit, overcomes the bottleneck of traditional blind selection of negative samples. Through semi-supervised learning, also known as the spying method, it effectively eliminates deceptively hidden "pseudo-negative samples," improving the quality of the dataset from the source and thus significantly enhancing the classification accuracy of the evaluation model. Specifically, in this embodiment, based on the positional relationship between the slope unit and existing landslide areas that have already experienced landslides, the slope units assigned multi-dimensional feature values ​​are divided into positive samples and a first unlabeled sample pool. That is, based on the pre-constructed landslide spatial database, slope units falling within the landslide boundary are labeled as positive samples (known landslide units), and the remaining slope units within the slope area that have not experienced known landslides are collectively labeled as the "unlabeled sample pool" (i.e., candidate non-landslide units, the first unlabeled sample pool). Subsequently, the system uses a semi-supervised strategy based on the spying method to process the positive samples and the first unlabeled sample pool to form training samples.

[0043] Furthermore, such as Figure 3 As shown, the step of using a spying method to process the positive samples and the first unlabeled samples to form the training samples includes: S403: Extract a portion of the positive samples as spy samples, and mix the spy samples into the first unlabeled samples to form a second unlabeled sample; S404: The training samples are formed based on the positive samples and a second unlabeled sample mixed with the spy samples.

[0044] For example, a set proportion, such as 10%, can be randomly selected from the known positive samples (landslide units) as spy samples. Their landslide labels are then removed, and these samples are mixed into the first unlabeled sample pool mentioned above to form the second unlabeled sample pool. The remaining positive samples and all unlabeled samples mixed with spy samples, i.e., the second unlabeled samples, are used as training data.

[0045] Then, the system can combine the environmental factor feature vectors corresponding to each unit to construct a binary classification algorithm model. For example, it can, but is not limited to, configure the initial prediction model as a logistic regression model with a regularization parameter of 0.1.

[0046] Specifically, the core logic of this embodiment is to evaluate the degree of difference between unlabeled units and the known landslide feature space, and identify the unlabeled sample with the highest heterogeneity to the landslide unit in the feature space as the true negative sample, that is, as the non-landslide area. In this embodiment, the step of calling the target prediction model to predict the slope unit that has not yet experienced a landslide includes: S601: Invoke the target prediction model to predict the spy sample and the first unlabeled sample; The method further includes: S7: Determine the predicted probability of the spy sample output by the target prediction model; S8: Determine the confidence threshold based on the predicted probability of the spy sample; S9: Compare the predicted probability of each of the first unlabeled samples based on the confidence threshold, and determine the non-landslide area based on the comparison result.

[0047] Specifically, the predicted probability of each of the first unlabeled samples is compared based on the confidence threshold, and non-slide areas are determined based on the comparison results, including: S901: Determine the slope unit corresponding to the first unlabeled sample whose predicted probability is lower than the confidence threshold as a non-slope unit; S902: All the aforementioned non-slope units constitute the non-slope area.

[0048] For example, the second unlabeled sample (i.e., the spy sample) and the first unlabeled sample are both input into the target prediction model to calculate the predicted probability of a landslide for each spy sample and the first unlabeled sample. Next, the predicted probability distribution characteristics of all spy samples are extracted and statistically analyzed, and a very low quantile of this distribution, such as, but not limited to, below 5%, is selected as the confidence threshold for determining non-landslide areas. Then, the predicted probability of the first unlabeled sample is compared with the confidence threshold. Slope units below this confidence threshold are determined as highly reliable, truly stable units, i.e., geologically stable, non-landslide areas that are unlikely to experience landslides. These non-landslide areas can also be used as purified non-landslide negative samples for other applications.

[0049] In another embodiment, to make the target prediction model interpretable, interpretability analysis of landslide disaster-causing patterns can be implemented, but is not limited to, according to the SHAP framework. Specifically, the target prediction model can be input into the SHAP (SHapleyAdditive exPlanations) interpreter module for interpretation as follows: Global interpretation: Quantify the overall marginal contribution of each environmental factor to the final output of the model, and rank the importance of the output features.

[0050] Local interpretation: Single-sample analysis of slope units with different susceptibility levels reveals the dominant causes and contributions of factors such as micro-topography, hydrology, or human activities to landslide risk within specific slope unit areas.

[0051] Based on the above embodiments, it is evident that the proposed solution significantly improves sample purity. Within the established slope unit framework, it overcomes the bottleneck of traditional blind selection of negative samples, effectively eliminating hidden "pseudo-negative samples" through semi-supervised learning, thereby improving dataset quality from the source and significantly enhancing the classification accuracy of the prediction model. Furthermore, the target prediction model in this application exhibits statistically significant robustness. Multiple random resampling mechanisms and dual verification using mean / standard deviation confirm that the method completely eliminates the random interference caused by a single dataset split, ensuring that the selected evaluation framework possesses strong generalization ability and stability in practical applications. In addition, the method in this application also achieves visualized decoding of the evaluation results. By introducing the SHAP module, not only can the predicted probability of the slope unit be obtained, but the prediction logic can also be understood, allowing for precise identification of the controlling factors inducing instability in specific slope units, providing more targeted scientific guidance for disaster prevention and mitigation.

[0052] like Figure 4 As shown, another embodiment of the present invention also provides a landslide susceptibility prediction device, comprising: The module is used to obtain basic data of the slope area, which is related to the topography, hydrology, and human engineering of the slope area. The division module is used to divide the ground surface of the slope area into multiple slope units with closed boundaries; The determination module is used to determine the multidimensional feature values ​​of each slope unit based on the basic data, wherein the multidimensional feature values ​​are related to the topography, hydrology, and human engineering of the slope unit; The preparation module is used to prepare multiple slope units containing the multidimensional feature values ​​to form training samples based on the positional relationship between the slope unit and the landslide area where a landslide has occurred. The training module is used to train an initial prediction model by combining the training samples to obtain a target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit. The prediction module is used to call the target prediction model to predict the slope units that have not yet experienced a landslide.

[0053] In one embodiment, the step of preparing training samples from multiple slope units containing the multidimensional feature values ​​based on the positional relationship between the slope unit and the landslide area where a landslide has occurred includes: Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are divided into positive samples and a first unlabeled sample. The positive sample indicates that the slope area is located within the landslide area, and the first unlabeled sample indicates that the slope area is located outside the landslide area. The training samples are formed by processing the positive samples and the first unlabeled samples using a spying method.

[0054] In one embodiment, the step of using a spying method to process the positive samples and the first unlabeled samples to form the training samples includes: A portion of the positive samples is extracted as spy samples, and the spy samples are mixed into the first unlabeled samples to form a second unlabeled sample; The training samples are formed based on the positive samples and a second unlabeled sample mixed with the spy samples.

[0055] In one embodiment, the step of calling the target prediction model to predict the slope units that have not yet experienced a landslide includes: The target prediction model is invoked to predict the spy sample and the first unlabeled sample; The device further includes: The first determining module is used to determine the predicted probability of the spy sample output by the target prediction model; The second determining module is used to determine a confidence threshold based on the predicted probability of the spy sample; The comparison module is used to compare the predicted probability of each of the first unlabeled samples based on the confidence threshold, and to determine the non-landslide area based on the comparison result.

[0056] In one embodiment, the predicted probability of each of the first unlabeled samples is compared based on the confidence threshold, and non-slide areas are determined based on the comparison results, including: The slope unit corresponding to the first unlabeled sample whose predicted probability is lower than the confidence threshold is determined as a non-slope unit; All the aforementioned non-landslide units constitute the non-landslide area.

[0057] In one embodiment, obtaining the basic data of the slope region includes: Obtain basic topographic data of the slope area, including slope, aspect, relative elevation difference, elevation, and landform type; Obtain hydrological distribution maps, including normalized water index; Obtain data on human engineering activities, including normalized difference vegetation index, normalized difference building index, land use type, and distance from roads.

[0058] In one embodiment, dividing the surface of the slope region into multiple closed-boundary slope units includes: Determine the DEM data of the slope area; The DEM data of the slope region is processed based on the terrain spatial analysis algorithm to adaptively generate multiple slope units with closed boundaries.

[0059] In one embodiment, determining the multidimensional feature values ​​of each slope unit based on the basic data includes: Determine the basic data corresponding to each slope unit; Spatial overlay analysis and feature extraction are performed on the basic data of the slope units to obtain multidimensional feature values ​​for each slope unit. The multidimensional feature values ​​include slope, aspect, relative elevation difference, elevation, landform type, normalized vegetation index, normalized water index, normalized building index, land use type, and distance from the road.

[0060] Another embodiment of the present invention also provides an electronic device, comprising: One or more processors; Memory, configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the landslide susceptibility prediction method as described above.

[0061] Furthermore, one embodiment of the present invention also provides a storage medium storing a computer program, which, when executed by a processor, implements the landslide susceptibility prediction method as described above. It should be understood that the various solutions in this embodiment have the corresponding technical effects in the above-described method embodiments, and will not be repeated here.

[0062] Furthermore, embodiments of the present invention also provide a computer program product tangibly stored on a computer-readable medium and comprising computer-readable instructions that, when executed, cause at least one processor to perform a landslide susceptibility prediction method such as those described in the embodiments above.

[0063] It should be noted that the computer storage medium of the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access storage medium (RAM), a read-only storage medium (ROM), an erasable programmable read-only storage medium (EPROM or flash memory), an optical fiber, a portable compact disk read-only storage medium (CD-ROM), an optical storage medium, a magnetic storage medium, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program configured for use by or in connection with an instruction execution system, system, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, antenna, optical fiber, RF, etc., or any suitable combination thereof.

[0064] Furthermore, those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0065] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0066] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

Claims

1. A method for predicting landslide susceptibility, characterized in that, include: Obtain basic data for the slope area, which is related to the topography, hydrology, and human engineering of the slope area; The surface of the slope region is divided into multiple closed-bound slope units; Based on the aforementioned basic data, multidimensional feature values ​​are determined for each slope unit. These multidimensional feature values ​​are related to the topography, hydrology, and human engineering of the slope unit. Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are prepared to form training samples. The initial prediction model is trained using the training samples to obtain the target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit. The target prediction model is invoked to predict the slope units that have not yet experienced a landslide.

2. The landslide susceptibility prediction method according to claim 1, characterized in that, The step of preparing training samples from multiple slope units containing the multidimensional feature values ​​based on the positional relationship between the slope unit and the landslide area where a landslide has occurred includes: Based on the positional relationship between the slope unit and the landslide area where a landslide has occurred, multiple slope units containing the multidimensional feature values ​​are divided into positive samples and a first unlabeled sample. The positive sample indicates that the slope area is located within the landslide area, and the first unlabeled sample indicates that the slope area is located outside the landslide area. The training samples are formed by processing the positive samples and the first unlabeled samples using a spying method.

3. The landslide susceptibility prediction method according to claim 2, characterized in that, The step of using a spying method to process the positive samples and the first unlabeled samples to form the training samples includes: A portion of the positive samples is extracted as spy samples, and the spy samples are mixed into the first unlabeled samples to form a second unlabeled sample; The training samples are formed based on the positive samples and a second unlabeled sample mixed with the spy samples.

4. The landslide susceptibility prediction method according to claim 3, characterized in that, The step of calling the target prediction model to predict the slope units that have not yet experienced a landslide includes: The target prediction model is invoked to predict the spy sample and the first unlabeled sample; The method further includes: Determine the predicted probability of the spy sample output by the target prediction model; The confidence threshold is determined based on the predicted probability of the spy sample; The predicted probability of each of the first unlabeled samples is compared based on the confidence threshold, and the non-landslide area is determined based on the comparison result.

5. The landslide susceptibility prediction method according to claim 4, characterized in that, The predicted probabilities of each of the first unlabeled samples are compared based on the confidence threshold, and non-slide areas are determined based on the comparison results, including: The slope unit corresponding to the first unlabeled sample whose predicted probability is lower than the confidence threshold is determined as a non-slope unit; All the aforementioned non-landslide units constitute the non-landslide area.

6. The landslide susceptibility prediction method according to claim 4, characterized in that, The basic data for obtaining the slope area includes: Obtain basic topographic data of the slope area, including slope, aspect, relative elevation difference, elevation, and landform type; Obtain hydrological distribution maps, including normalized water index; Obtain data on human engineering activities, including normalized difference vegetation index, normalized difference building index, land use type, and distance from roads.

7. The landslide susceptibility prediction method according to claim 1, characterized in that, The process of dividing the surface of the slope region into multiple closed-bound slope units includes: Determine the DEM data of the slope area; The DEM data of the slope region is processed based on the terrain spatial analysis algorithm to adaptively generate multiple slope units with closed boundaries.

8. The landslide susceptibility prediction method according to claim 1, characterized in that, The determination of the multidimensional feature values ​​of each slope unit based on the basic data includes: Determine the basic data corresponding to each slope unit; Spatial overlay analysis and feature extraction are performed on the basic data of the slope units to obtain multidimensional feature values ​​for each slope unit. The multidimensional feature values ​​include slope, aspect, relative elevation difference, elevation, landform type, normalized vegetation index, normalized water index, normalized building index, land use type, and distance from the road.

9. A landslide susceptibility prediction device, characterized in that, include: The module is used to obtain basic data of the slope area, which is related to the topography, hydrology, and human engineering of the slope area. The dividing module is used to divide the ground surface of the slope area into multiple slope units with closed boundaries; The determination module is used to determine the multidimensional feature values ​​of each slope unit based on the basic data, wherein the multidimensional feature values ​​are related to the topography, hydrology, and human engineering of the slope unit; The preparation module is used to prepare multiple slope units containing the multidimensional feature values ​​to form training samples based on the positional relationship between the slope unit and the landslide area where a landslide has occurred. The training module is used to train an initial prediction model by combining the training samples to obtain a target prediction model, which is used to predict the probability of a landslide occurring in the input slope unit. The prediction module is used to call the target prediction model to predict the slope units that have not yet experienced a landslide.

10. An electronic device, comprising: One or more processors; Memory, configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the landslide susceptibility prediction method as described in any one of claims 1-8.

Images