Landslide susceptibility prediction method, system and equipment and storage medium

Abstract

The embodiment of the invention provides a landslide susceptibility prediction method, system and device and a computer readable storage medium. The method comprises the steps of obtaining a to-be-divided topographic image; wherein the topographic image is a slope direction and mountain shadow map extracted based on the digital topographic image; performing region division on the topographic image through a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used for carrying out slope unit division on the terrain; based on the slope unit, environment factor extraction is carried out, and multiple evaluation factors are determined; performing landslide susceptibility prediction on the slope unit based on the plurality of evaluation factors through a landslide susceptibility prediction model to obtain a landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide susceptibility prediction model comprises a self-organizing mapping neural network module, an information amount module and a support vector machine module; the landslide susceptibility prediction model is a model which is trained by adopting historical landslide data and is used for carrying out landslide prediction on the slope unit. According to the scheme, the accuracy of landslide susceptibility prediction can be improved.

Description

Technical Field

[0001] The present application relates to the field of geographic information technology, and in particular to a landslide susceptibility prediction method, system, device and computer-readable storage medium. Background Art

[0002] Geological hazards are disastrous geological phenomena caused by various geological processes. They are influenced by both the natural environment and human activities and are characterized by complex causes, widespread occurrence, and recurrence. Landslides are one of the most frequent and most devastating types of geological hazards, posing a serious threat to the safety of local residents and their property. Geological hazard susceptibility assessments, based on geological environmental conditions and combining factors that predispose and induce geological hazards, predict the spatial probability of potential landslides. Regional landslide susceptibility assessment models have become a key component of disaster research.

[0003] In recent years, domestic and foreign scholars have conducted a series of related studies on the quantitative prediction of landslide disasters. Although these methods have improved the accuracy and efficiency of spatial prediction of landslide disasters to a certain extent, the selection of landslide evaluation units and the construction of prediction models still have defects, resulting in low accuracy of landslide susceptibility prediction. Summary of the Invention

[0004] The embodiments of the present application are intended to provide a landslide susceptibility prediction method, system, device, and computer-readable storage medium that can improve the accuracy of landslide susceptibility prediction.

[0005] The technical solution of the present invention is achieved as follows:

[0006] In a first aspect, an embodiment of the present application provides a method for slope unit division and landslide susceptibility prediction, the method comprising:

[0007] Acquire a terrain image to be divided; wherein the terrain image is a slope aspect and hill shadow map extracted based on a digital terrain image;

[0008] Dividing the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units;

[0009] Based on the slope unit, environmental factors are extracted to determine multiple evaluation factors;

[0010] Through the landslide susceptibility prediction model, the landslide susceptibility of the slope unit is predicted based on the multiple evaluation factors to obtain the landslide susceptibility prediction results corresponding to the slope unit; wherein, the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to predict landslides for slope units.

[0011] In the above solution, the terrain image is divided into regions using a multi-scale algorithm to obtain slope units, including:

[0012] Dividing the terrain image into regions using the multi-scale algorithm to obtain multiple image regions;

[0013] Based on the minimum heterogeneity principle, any two adjacent image areas in the multiple image areas are merged to determine an initial slope unit;

[0014] The initial slope units are segmented, merged and smoothed to determine the slope units.

[0015] In the above solution, the merging of any two adjacent image regions among the multiple image regions based on the minimum heterogeneity principle to determine the initial slope unit includes:

[0016] Based on the minimum heterogeneity principle, any two adjacent image regions among the multiple image regions are merged to obtain a target image region; wherein the area of ​​the target image region is larger than the area of ​​the image region;

[0017] calculating the heterogeneity of the target image region;

[0018] If the heterogeneity is greater than the scale threshold, the two adjacent image regions are not merged, and the two adjacent image regions are respectively determined as the initial slope units;

[0019] If the heterogeneity is less than the scale threshold, the merging process is continued based on the target image region until the heterogeneity of the obtained new target image region is greater than the scale threshold, and the initial slope unit is determined.

[0020] In the above solution, before dividing the terrain image into regions using a multi-scale algorithm to obtain slope units, the method further includes:

[0021] Acquiring historical landslide data of a study area corresponding to the terrain image to be divided;

[0022] Calculating landslide area and shape characteristics based on the historical landslide data;

[0023] Based on the landslide area and the shape characteristics, multiple parameter adjustments are performed to determine the optimal parameters of the multi-scale algorithm; wherein the optimal parameters include scale, shape feature weight and compactness weight.

[0024] In the above solution, the landslide susceptibility prediction model is used to predict the landslide susceptibility of the slope unit based on the multiple evaluation factors to obtain the landslide susceptibility prediction result corresponding to the slope unit, including:

[0025] Based on the multiple evaluation factors, performing initial susceptibility differentiation through the self-organizing map neural network module to determine an initial susceptibility evaluation result;

[0026] By means of the information quantity module, the plurality of evaluation factors are graded and information quantity values ​​are calculated to determine the information quantity values ​​corresponding to the plurality of evaluation factors;

[0027] The support vector machine module is used to predict the landslide susceptibility of the slope unit based on the information amount values ​​corresponding to the multiple evaluation factors, so as to obtain the landslide susceptibility prediction result corresponding to the slope unit.

[0028] In the above solution, the step of grading the multiple evaluation factors and calculating the information quantity values ​​by the information quantity module to determine the information quantity values ​​corresponding to the multiple evaluation factors includes:

[0029] Using the information volume module, grading each of the multiple evaluation factors to obtain a secondary classification result of each evaluation factor;

[0030] Calculating the information value of each evaluation factor based on the secondary classification result of each evaluation factor and the probability of geological disasters occurring in the study area corresponding to the acquired terrain image;

[0031] Based on the information amount value of each evaluation factor, the information amount values ​​corresponding to each of the plurality of evaluation factors are determined.

[0032] In the above solution, before performing landslide susceptibility prediction on the slope unit based on the multiple evaluation factors using the landslide susceptibility prediction model and obtaining the landslide susceptibility prediction result corresponding to the slope unit, the method further includes:

[0033] Acquire landslide sample data in the study area; wherein the landslide sample data is the slope unit where the historical landslide is located;

[0034] Determining a plurality of environmental assessment factors based on the study area;

[0035] Based on the multiple environmental assessment factors, performing initial susceptibility differentiation and non-landslide sample production through a self-organizing map neural network module to determine non-landslide sample data;

[0036] Determining training sample data and test sample data based on the landslide sample data and the non-landslide sample data;

[0037] Training the initial landslide susceptibility prediction model using the training sample data to obtain a trained landslide susceptibility prediction model;

[0038] The trained landslide susceptibility prediction model is tested using the test sample data to obtain a test result; and when the test result is greater than a preset error threshold, the landslide susceptibility prediction model is determined.

[0039] In a second aspect, an embodiment of the present application provides a landslide susceptibility prediction system, comprising: an acquisition unit, a division unit, a determination unit, and a prediction unit, wherein:

[0040] The acquisition unit is used to acquire a terrain image to be divided; wherein the terrain image is a slope aspect and hill shadow map extracted based on a digital terrain image;

[0041] The division unit is used to divide the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units;

[0042] The determining unit is configured to extract environmental factors based on the slope unit and determine a plurality of evaluation factors;

[0043] The prediction unit is used to predict the landslide susceptibility of the slope unit based on the multiple evaluation factors through a landslide susceptibility prediction model to obtain a landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to predict landslides on slope units.

[0044] In a third aspect, an embodiment of the present application provides a landslide susceptibility prediction device, the landslide susceptibility prediction device comprising: a processor and a memory; wherein,

[0045] The memory is used to store computer programs;

[0046] The processor is configured to call and run the computer program from the memory to execute the method according to the first aspect.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing executable instructions for causing a processor to execute the method described in the first aspect.

[0048] An embodiment of the present application provides a landslide susceptibility prediction method, system, device and computer-readable storage medium, the method comprising: obtaining a terrain image to be divided; wherein the terrain image is a slope aspect and mountain shadow map extracted based on a digital terrain image; performing regional division on the terrain image using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units; based on the slope units, environmental factors are extracted to determine multiple evaluation factors; using a landslide susceptibility prediction model, landslide susceptibility prediction is performed on the slope units based on the multiple evaluation factors to obtain a landslide susceptibility prediction result corresponding to the slope units; wherein the landslide susceptibility prediction model comprises: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to perform landslide prediction on slope units. In the above scheme, on the one hand, a multi-scale algorithm is used to divide the terrain image into regions to obtain slope units. Since the multi-scale algorithm performs regional division based on the extracted slope aspect and hillshade map, the accuracy of slope unit division can be improved. On the other hand, since the landslide susceptibility prediction model is a pre-trained landslide prediction model and includes a self-organizing map neural network module, an information content module, and a support vector machine module, the self-organizing map neural network module can classify the landslide susceptibility based on the slope units, the information content module can calculate the information content values ​​of multiple evaluation factors, and the support vector machine module can predict the landslide susceptibility of the slope units based on the information content values ​​of multiple evaluation factors. In this process, the integration of the three modules gradually distinguishes and calculates the obtained slope units to achieve landslide susceptibility prediction, thereby improving the accuracy of the final landslide susceptibility prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings herein are incorporated into and constitute a part of this specification. These drawings illustrate embodiments consistent with the present application and, together with the specification, serve to illustrate the technical solutions of the present application. Obviously, the drawings described below are merely some embodiments of the present application. Those skilled in the art can, without inventive effort, derive other drawings from these drawings.

[0050] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0051] Figure 1 An optional process diagram of a landslide susceptibility prediction method is provided for the embodiment of the present application Figure 1 ;

[0052] Figure 2 A schematic diagram of multiple evaluation factors of a landslide susceptibility prediction method is provided for an embodiment of the present application;

[0053] Figure 3 An optional process diagram of a landslide susceptibility prediction method is provided for the embodiment of the present application Figure 2

[0054] Figure 4 An optional process diagram of a landslide susceptibility prediction method is provided for the embodiment of the present application Figure 3 ;

[0055] Figure 5 A schematic structural diagram of a landslide susceptibility prediction system is provided for an embodiment of the present application;

[0056] Figure 6 A structural schematic diagram of a landslide susceptibility prediction device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0057] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the specific technical solutions of the present application will be further described in detail below in conjunction with the drawings in the embodiments of the present application. The following embodiments are used to illustrate the present application but are not intended to limit the scope of the present application.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0059] In the following description, references to “some embodiments,” “this embodiment,” “embodiments of the present application,” and examples, etc., describe a subset of all possible embodiments. However, it can be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict.

[0060] If similar descriptions of "first / second" appear in the application documents, the following explanation is added. In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0061] Based on this, the present invention provides a method for predicting landslide susceptibility. Figure 1 An optional process diagram of a landslide susceptibility prediction method is provided for the embodiment of the present application Figure 1 , will combine Figure 1 The steps shown are explained.

[0062] S101. Obtain a terrain image to be divided; wherein the terrain image is a slope aspect and mountain shadow map extracted based on a digital terrain image.

[0063] In the embodiments of the present application, a hillshade map is a technique for determining terrain based on the slope and aspect of a light source and elevation surface. It can reflect certain terrain information, and its image grayscale value ranges from 0 to 255. The aspect is defined as the direction of the slope normal projected onto the horizontal plane, and can be used to identify the steepest descent direction at a location on the ground. Therefore, based on the characteristics of the aspect and terrain shadows, a multi-scale algorithm can subsequently be used to maximize the uniformity within the slope unit and minimize the heterogeneity between slope units.

[0064] In some embodiments of the present application, the slope unit division and landslide susceptibility prediction method is applicable to scenarios where geological environment data is extracted for various terrains.

[0065] In some embodiments of the present application, the slope unit division and landslide susceptibility prediction method is executed by a slope unit division and landslide susceptibility prediction device. The slope unit division and landslide susceptibility prediction device can be a server or a terminal device, which is not specifically limited in the embodiments of the present application.

[0066] In some embodiments of the present application, the slope unit division and landslide susceptibility prediction device can obtain an initial terrain image and extract shadow maps of each mountain in the slope direction from the initial terrain image.

[0067] It should be noted that the initial terrain image is a terrain map of original appearance directly captured by an image acquisition device.

[0068] For example, let's take the location of XX City in XX Province. Its geographic coordinates lie between 104°11' and 104°57' east longitude and 25°19' and 25°57' north latitude, covering a total area of ​​4,057 square kilometers. Located on a sloping slope where the Yunnan-Guizhou Plateau transitions to the hilly plains of Guangxi, Panzhou's terrain is generally characterized by high terrain in the north and low terrain in the south, with high terrain on the east and west sides and low terrain in the center. The surface features mountains, hills, and basins, with steep terrain and steep slopes, and numerous fault zones and fold belts. The region has a subtropical humid climate, with an average annual temperature of approximately 15.2 degrees Celsius and an average annual precipitation of approximately 1,390 mm. This humid climate, combined with complex landforms, diverse stratigraphic structures, and frequent tectonic activity, makes Panzhou's geological environment fragile, particularly prone to geological disasters during the summer rainy season. According to the National Geological Hazard Prevention and Control Plan, the entire Panzhou area is classified as a geological disaster-prone area. Types of geological disasters include landslides, collapses, mudslides, ground subsidence and ground fissures, among which landslide disasters are more prominent. Through data collection and investigation, it is known that there have been 160 landslide disasters in the area since 2016, accounting for 67.45% of the total number of geological disasters. From the perspective of spatial distribution, geological disasters are distributed to varying degrees in the 31 townships in the area, among which the distribution is more concentrated in the northern and central regions, and relatively sparse in the southern region.

[0069] S102 , dividing the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units.

[0070] In the embodiments of this application, the shaping process of the slope unit is associated with the natural morphology and landforms, thereby reflecting the physical relationship between the landslide and geological environment information. The acquisition and application of slope units are receiving increasing attention. According to traditional hydrological principles, a slope unit is defined as the intersection of the ridgeline of the drainage area and the valley line of the catchment area. The ridgeline is the boundary of the watershed, and the area between the two ridgelines is the catchment area. Extracting the valley line and segmenting the watershed can obtain two slope units.

[0071] In some embodiments of this application, from a geomorphological perspective, a slope unit corresponds not only to a single slope but also to multiple slopes or even an entire watershed. In basic hydrological analysis, new definitions and restrictions have been added to slope units. Specifically, a slope unit is considered an area with significantly different topographic features from adjacent areas. In addition to ridgelines and valley lines, topographic discontinuities such as elevation, slope, and curvature are also considered as demarcation lines.

[0072] In some embodiments of the present application, a terrain image is segmented using a multi-scale algorithm to obtain multiple image regions. Based on the principle of minimum heterogeneity, any two adjacent image regions in the multiple image regions are merged to determine initial slope units. The initial slope units are segmented, merged, and smoothed to determine slope units.

[0073] In some embodiments of the present application, before performing regional segmentation of the terrain image using a multi-scale algorithm to obtain slope units, the slope unit segmentation and landslide susceptibility prediction method further includes:

[0074] Obtain historical landslide data of the study area corresponding to the terrain image to be divided;

[0075] Calculate landslide area and shape characteristics based on historical landslide data;

[0076] Based on the landslide area and shape characteristics, multiple parameter adjustments were performed to determine the optimal parameters of the multi-scale algorithm; among them, the optimal parameters include scale, shape feature weight and compactness weight.

[0077] For example, parameter selection directly determines the division result of slope units. The parameters mainly include scale, shape feature weight and compactness weight. If the scale parameter is too large, incomplete segmentation will occur; otherwise, fragmented segmentation will occur. The best parameter combination is selected by the modified trial and error method. The area and shape characteristics of the landslide in the study area are calculated. The shape characteristics are represented by the shape index (R value). The R value can be obtained by calculating the formula (R = L^2 / (S), L and S are the perimeter and area of ​​the slope surface respectively); the parameters that need to be set in the MSS method include scale, shape feature weight and compactness weight parameters, which are determined by the trial and error method and combined with the morphological and scale characteristics of historical landslides in the study area (modified trial and error method).

[0078] S103. Extract environmental factors based on the slope unit and determine multiple evaluation factors.

[0079] In some embodiments of the present application, the multiple evaluation factors include: terrain relief, slope, aspect, average annual rainfall, distance to structure, distance to water system, distance to road and lithology.

[0080] Topographic relief: Relative elevation is the difference between the highest and lowest points in a cell, which determines the magnitude of slope stress. The greater the relative elevation, the greater the likelihood of a landslide. The focal statistics tool in ArcGIS Surface Analysis calculates the relative elevation within a region, or topographic relief, and categorizes the calculated elevation into four levels.

[0081] Slope: Slope intuitively describes the undulating shape of a local surface. Steeper slopes increase the intensity of surface material flow and energy conversion, potentially reducing slope stability and thus affecting the occurrence of disasters. Based on DEM data, the ArcGIS Slope tool was used to generate a slope distribution map, which was divided into six levels.

[0082] Slope aspect: Slope aspect refers to the direction of the slope normal projected onto the horizontal plane. Different slopes experience different natural conditions, such as solar radiation intensity and rainfall, leading to differences in vegetation characteristics and rock weathering. Therefore, slope aspect is a key factor influencing the distribution of geological hazards. Using the ArcGIS Neighborhood Analysis spatial analysis tool, the surface tool categorizes slope aspect into nine levels: southeast, northwest, northeast, southeast, southwest, northwest, and flat.

[0083] Lithology: Different types of engineering geological rock groups have significantly different effects on the formation of landslide geological hazards. They are divided into four rock groups based on the hardness of the rock, the purity of the components and the weak interlayers: ① Loose soil; ② Harder to softer thin-thick layered granulite and quartz schist interbedded rock group; ③ Harder medium-thick layered granulite rock group; ④ Hard massive metadiabase rock group.

[0084] Average annual rainfall: Rainfall not only affects the soil moisture content, but also causes the slope structure to become unbalanced after the surface soil layer is eroded, thereby causing geological disasters. The rainfall results are divided into four levels based on the natural breakpoint method.

[0085] Distance from the water system: The water system mainly causes erosion and damage to the river bank. Combined with the development characteristics of landslide geological hazards in the study area, the buffer tool of ArcGIS neighborhood analysis was used to conduct a buffer analysis on the water system to obtain the river influence range, and the water system influence range was divided into four levels.

[0086] Distance from roads: The road factor primarily reflects the impact of human engineering activities. Road construction inevitably involves cut-and-fill reconstruction, which can easily lead to steep slopes. Under the long-term effects of gravity and external forces, the original slope structure is destroyed, creating conditions for geological disasters. To standardize data, the buffer tool in ArcGIS was used to divide the distance from roads into six equal levels at 400m intervals.

[0087] Distance from structures: Faults primarily affect the rock and soil structure within a certain range, reducing slope stability and providing groundwater access, which in turn causes slope deformation. Taking into account the spatial distribution relationship between faults and geological hazards in the study area, the buffer zone tool in ArcGIS Neighborhood Analysis was used to divide the faults into four segments at 1000m intervals.

[0088] In some embodiments of the present application, environmental factors are extracted based on the specific characteristics of the slope unit and the study area to determine multiple evaluation factors.

[0089] For example, the occurrence of landslide disasters is affected by multiple temporal and spatial factors. Based on the specific characteristics of the study area and existing research results, the development characteristics of geological hazards in XX City and the spatial distribution of influencing factors were analyzed in this study area. After comprehensive consideration of the availability of data, the scale and scope of the study area, the following factors were finally selected: Figure 2 Eight factors were evaluated: elevation (i.e., relief), slope (i.e., slope), aspect (i.e., aspect), annual average rainfall (i.e., annual average rainfall), distance to faults (i.e., distance to structures), distance to drainage (i.e., distance to drainage), distance to roads (i.e., distance to roads), and lithologic distribution (i.e., lithologic). Considering the heterogeneity of environmental factors within slope units, each environmental factor can be characterized using three variables: mean, variance, and standard deviation. It should be noted that discrete variables (lithologic and aspect) are represented using the mode, making it impossible to obtain a similar three-variable representation.

[0090] S104. Using a landslide susceptibility prediction model, perform landslide susceptibility prediction on the slope unit based on multiple evaluation factors to obtain a landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module, and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to predict landslides on slope units.

[0091] In some embodiments of the present application, the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a SOM-I-SVM model.

[0092] In some embodiments of the present application, the landslide susceptibility prediction device can use a landslide susceptibility prediction model to predict the landslide susceptibility of a slope unit through multiple evaluation factors, obtain a prediction value corresponding to the slope unit, and determine the landslide susceptibility prediction result through the prediction value; the landslide susceptibility prediction result includes landslide or non-landslide.

[0093] It should be noted that the predicted value can be 0 or 1; 1 indicates landslide, and 0 indicates non-landslide.

[0094] For example, all 164,827 slope units in the entire region were converted into surface elements, and the evaluation factors and their information values ​​were extracted and brought into the trained model to calculate the landslide susceptibility index. The natural breakpoint method was used to divide the susceptibility of landslide disasters into five levels: extremely high susceptibility area, high susceptibility area, medium susceptibility area, low susceptibility area, and extremely low susceptibility area. The results of susceptibility zoning and the distribution number of landslide points of each grade were statistically analyzed. Among them, the support vector machine (SVM) model showed that the extremely high susceptibility areas accounted for 14.6%, the high susceptibility areas accounted for 18.3%, the medium susceptibility areas accounted for 26.4%, the low susceptibility areas accounted for 22.3%, and the low susceptibility areas accounted for 18.4%; the support vector machine model based on information quantity (I-SVM) showed that the extremely high susceptibility areas accounted for 13.0%, the high susceptibility areas accounted for 17.0%, the medium susceptibility areas accounted for 21.7%, the low susceptibility areas accounted for 2.9%, and the low susceptibility areas accounted for 25.4%; the support vector machine model based on self-organizing map neural network and information quantity (SOM-I-SVM) showed that the extremely high susceptibility areas accounted for 12.7%, the high susceptibility areas accounted for 17.0%, the medium susceptibility areas accounted for 20.7%, the low susceptibility areas accounted for 24.1%, and the low susceptibility areas accounted for 25.5%. Table 1 shows the parameter settings and model scores for the SVM, I-SVM, and SOM-I-SVM models. A multi-scale segmentation algorithm, combined with a trial-and-error approach to determine the optimal parameter combination for the MSS, enabled a precise and efficient segmentation of slope units. An evaluation index system was then established based on eight influencing factors, including elevation difference, slope gradient, aspect, annual average rainfall, and distance to water systems, selected from the slope units. This was then combined with the SOM neural network to conduct a preliminary regional susceptibility assessment. Non-landslide samples were selected from low-susceptibility areas and evaluated for landslide susceptibility using the SVM, I-SVM, and SOM-I-SVM models. Compared to the SVM and I-SVM models, the SOM-I-SVM model achieved the best susceptibility assessment results, with accuracy improvements of 12% and 4.7%, respectively.

[0095] Table 1

[0096]

[0097] It can be understood that, on the one hand, the terrain image is regionalized using a multi-scale algorithm to obtain slope units. Since the multi-scale algorithm performs regional division based on the extracted slope aspect and hillshade map, it can improve the accuracy of slope unit division. On the other hand, since the landslide susceptibility prediction model is a pre-trained landslide prediction model, and the landslide susceptibility prediction model includes a self-organizing map neural network module, an information content module, and a support vector machine module, the self-organizing map neural network module can classify the degree of landslide hazard susceptibility based on slope units, the information content module can calculate the information content values ​​of multiple evaluation factors, and the support vector machine module can predict the landslide susceptibility of slope units based on the information content values ​​of multiple evaluation factors. In this process, the integration of these three modules gradually distinguishes and calculates the obtained slope units to achieve landslide susceptibility prediction, thereby improving the accuracy of the final landslide susceptibility prediction.

[0098] In some embodiments of the present application, Figure 3 As shown, S102 can be implemented through S1021, S1022 and S1023 as follows:

[0099] S1021. Divide the terrain image into regions using a multi-scale algorithm to obtain multiple image regions.

[0100] S1022. Based on the minimum heterogeneity principle, any two adjacent image regions among the multiple image regions are merged to determine an initial slope unit.

[0101] In some embodiments of the present application, based on the principle of minimum heterogeneity, any two adjacent image regions among multiple image regions are merged to obtain a target image region; wherein the area of ​​the target image region is larger than the area of ​​the image region; the heterogeneity of the target image region is calculated; if the heterogeneity is greater than a scale threshold, the any two adjacent image regions are not merged, and the any two adjacent image regions are respectively determined as initial slope units; if the heterogeneity is less than a scale threshold, the merging process is continued based on the target image region until the heterogeneity of the obtained new target image region is greater than the scale threshold, and the initial slope unit is determined.

[0102] For example, based on a terrain image, a multi-scale algorithm is used starting from the pixel layer to first aggregate pixels with similar features in adjacent areas into small image areas (i.e., multiple image areas). Similar small image areas are then merged into a large image area (i.e., the target image area) based on the principle of minimum heterogeneity. Each time the two areas are merged, the heterogeneity of the merged area is calculated to see if it is greater than the scale threshold. If it is greater than the scale threshold, the two areas are not merged; if it is less than the scale threshold, the two areas are merged to generate a new, larger image area. The merging process stops until the heterogeneity of the merged areas is greater than the scale threshold, or all areas have been merged.

[0103] S1023: Segment, merge, and smooth the initial slope units to determine the slope units.

[0104] In some embodiments of the present application, the initial slope units are subjected to optimization processing such as segmentation, merging, and smoothing to eliminate objects with local "island effects" and finally obtain the slope units within the study area.

[0105] It can be understood that, through the multi-scale algorithm, the terrain image is divided into regions to obtain multiple image regions; due to the adoption of the minimum heterogeneity principle, any two adjacent image regions in the multiple image regions are merged to determine the initial slope unit, and the initial slope unit is segmented, merged and smoothed to determine the slope unit, so that the determined slope unit is more reasonable, which is convenient for the subsequent landslide susceptibility prediction based on the slope unit.

[0106] In some embodiments of the present application, Figure 4 As shown, S104 can be implemented through S1041, S1042 and S1043 as follows:

[0107] S1041. Based on multiple evaluation factors, perform initial susceptibility differentiation through a self-organizing map neural network module to determine an initial susceptibility evaluation result.

[0108] In some embodiments of the present application, the self-organizing neural network module is an unsupervised learning algorithm proposed by Kohonen in 1981. It performs self-organizing learning on the input data based on the topological relationship between the original data. The result of the learning is to obtain the winning neuron corresponding to the input data. The self-organizing neural network includes an input layer and an output layer. The input layer is mainly used for data input, and the number of its neurons is the same as the dimension of the input data. Its expression is

[0109] x ij =[X i1 X i2 ...X im ] (1)

[0110] Where: n is the total number of samples; m is the attribute dimension of sample data; X ij is the value of the jth attribute in the i-th sample, i = 1, 2, ..., n, j = 1, 2, ..., m.

[0111] In the output layer, neurons are arranged in a two-dimensional space in a certain pattern. Common arrangements include rectangles and hexagons. Before learning, each neuron is assigned an initial weight, which can be written as

[0112] w ij =[w i1 w i2 ...w im ] (2)

[0113] Where: N is the total number of neurons in the output layer; w ij is the weight value of the i-th neuron for the j-th attribute of the current input sample, i = 1, 2, ..., n.

[0114] The output layer is primarily used to output learning results. Neurons in the output layer are connected to their neighboring neurons through neighborhood relationships. In essence, a self-organizing neural network calculates the distance between neurons and input vectors based on their weights. The neuron with the smallest distance wins, and the input data is mapped to the winning neuron. The position of the winning neuron is continuously adjusted through iteration. After training, similar data points are mapped to neighboring neurons, thus achieving clustering.

[0115] The complexity of a self-organizing neural network is quadratically related to the number of neurons in the output layer. The rationality of the network size design is the key to the algorithm implementation. Generally speaking, the total number of neurons in the output layer is:

[0116]

[0117] Generally, the output layer network is composed of neurons with the same number of rows and columns (h), so the number of rows and columns of neurons is:

[0118]

[0119] S1042: Using the information quantity module, multiple evaluation factors are graded and information quantity values ​​are calculated to determine the information quantity values ​​corresponding to each of the multiple evaluation factors.

[0120] In some embodiments of the present application, each of the multiple evaluation factors is graded through an information quantity module to obtain a secondary classification result of each evaluation factor; based on the secondary classification result of each evaluation factor and the probability of geological disasters occurring in the study area corresponding to the acquired terrain image, the information quantity value of each evaluation factor is calculated; based on the information quantity value of each evaluation factor, the information quantity values ​​corresponding to each of the multiple evaluation factors are determined.

[0121] The information volume module, also known as the information volume model, is a statistical analysis and prediction method that reflects the susceptibility of geological disasters through the number of entropy values ​​at the time of geological disaster occurrence. By classifying each evaluation factor and using the frequency or density of geological disasters to reflect the magnitude of the secondary catastrophic effect under each evaluation factor, the actual data values ​​of the geological disaster evaluation factors are converted into information volume representing the magnitude of geological disaster susceptibility. By superimposing the information volume of each factor on the ArcGIS platform, the total information volume is calculated. The larger the total information volume value, the higher the geological disaster susceptibility.

[0122]

[0123] Where: I(A i-j )-the information content of the jth secondary classification of the i-th evaluation factor; P(A i-j )-the probability of occurrence of the jth secondary classification geological disaster of the ith evaluation factor; P(B)-the probability of occurrence of geological disasters in the study area.

[0124] In the actual calculation, since the scale of geological disaster points in the study area is small except for one medium-sized one, the sample frequency is used instead of the overall probability method in the single factor calculation, and formula (5) is converted into:

[0125]

[0126] Where: N i-j - the number of geological disasters occurring in the jth secondary classification of the i-th evaluation factor; s i-j - Area of ​​the jth secondary classification slope unit of the i-th evaluation factor, km 2 ; N- total number of geological hazards in the study area; S- total area of ​​the study area, km 2 .

[0127] The single factor information is superimposed and calculated to obtain the evaluation results of the multi-factor comprehensive information:

[0128]

[0129] Where: I(Ai) is the information content of the i-th evaluation factor; I is the comprehensive information content of multiple factors; n is the total number of evaluation factors; and frequency ratio is the ratio of the two factors. This model uses the total information content I as an indicator of the susceptibility of geological hazards within a slope unit. A larger value indicates a higher susceptibility to geological hazards within that slope unit.

[0130] S1043. Using a support vector machine module, based on the information values ​​corresponding to the multiple evaluation factors, predict the landslide susceptibility of the slope unit to obtain a landslide susceptibility prediction result corresponding to the slope unit.

[0131] In some embodiments of the present application, the support vector machine is based on machine learning theory and was first proposed by VAPNIK in the 1960s to study small sample cases. It is considered to be the best theory currently for small sample estimation and prediction learning, and is widely used in disaster susceptibility assessment.

[0132] The classification of data points can be achieved by calculating the distance between them and the hyperplane, and the credibility of the result is positively correlated with the distance. The objective function of this model can be defined as: maxγ has:

[0133]

[0134] By definition It can be seen that: Let γ equal to 1, the above formula becomes:

[0135]

[0136] This is equivalent to the corresponding constraint y i (w T x i +b)≥1, i=1, 2, 3..., n. maximize value, and This is the geometric margin γ. Support vector machines (SVMs) use kernel functions to solve nonlinear classification problems. Currently, the most commonly used SVM kernel functions include linear, polynomial, radial basis, and sigmoid kernels. The radial basis kernel function is widely used in disaster sensitivity assessment. However, for different research purposes, the prediction accuracy of the four kernel functions should be compared to select the optimal kernel function to establish an SVM prediction model.

[0137] It can be understood that since the landslide susceptibility prediction model is a landslide prediction model trained in advance, and the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module, the self-organizing map neural network module can divide the susceptibility of landslide disasters based on slope units, the information volume module can calculate the information volume values ​​of multiple evaluation factors, and the support vector machine module realizes landslide susceptibility prediction of slope units based on the information volume values ​​of multiple evaluation factors. In this process, due to the fusion of the three modules, the slope units are gradually differentiated and calculated to realize landslide susceptibility prediction, thereby improving the accuracy of the landslide susceptibility prediction.

[0138] In some embodiments of the present application, before executing S104, the slope unit division and landslide susceptibility prediction method further includes:

[0139] Obtain landslide sample data in the study area; landslide sample data refers to the slope units where historical landslides are located;

[0140] Determine multiple environmental assessment factors based on the study area;

[0141] Based on multiple environmental assessment factors, the self-organizing map neural network module is used to perform initial susceptibility differentiation and non-landslide sample production, and the non-landslide sample data is determined;

[0142] Based on the landslide sample data and the non-landslide sample data, determining the training sample data and the test sample data;

[0143] The initial landslide susceptibility prediction model is trained by using the training sample data to obtain a trained landslide susceptibility prediction model;

[0144] The trained landslide susceptibility prediction model is tested by testing sample data to obtain a test result; and when the test result is greater than a preset error threshold, the landslide susceptibility prediction model is determined.

[0145] For example, the information of the above eight environmental factors is used as the input variables of the machine learning model (landslide susceptibility prediction model), and the output variables of the model are 1 (landslide) and 0 (non-landslide). Among them, the input variables mainly include landslide samples (label value is 1) and non-landslide samples (label value is 0), which are divided into training set and test set according to a certain ratio, respectively used for training the machine learning model and testing the machine learning model. The landslide samples are the slope units where historical landslides are located, and the non-landslide samples are slope units randomly selected from the slope units outside the non-landslide area, with the same number of slope units as the landslide slope units. The landslide samples and non-landslide samples are divided into training set and test set in a ratio of 7:3.

[0146] In some embodiments of the present application, the SVM model can reflect the nonlinear relationship between various indicator factors. At the same time, the information volume can be used as a carrier to solve the problem of inconsistent factor dimensions. The advantages of the two models are fully combined to evaluate the susceptibility of landslide disasters. The steps of the SOM-I-SVM coupling model are mainly as follows:

[0147] (1) Based on the evaluation factors selected by correlation analysis, the SOM neural network is used to perform initial susceptibility partitioning and randomly generate non-landslide sample points in the low-susceptibility area;

[0148] (2) The evaluation factors are graded and the grade of each factor is calculated and used as the input variable to train the model, and the parameters are repeatedly adjusted to create the information value;

[0149] (3) The information value of each factor is used as the input of the initial SOM-I-SVM model, the initial SOM-I-SVM model is trained and the rationality is checked, the SOM-I-SVM model is determined, and the SOM-I-SVM model is used to predict landslide susceptibility.

[0150] It can be understood that by obtaining landslide sample data in the study area and determining non-landslide sample data, training sample data and test sample data are determined based on the landslide sample data and non-landslide sample data; the initial landslide susceptibility prediction model is trained through the training sample data to determine the landslide susceptibility prediction model, which can improve the accuracy of landslide susceptibility prediction of the landslide susceptibility prediction model.

[0151] Based on the landslide susceptibility prediction method of the above embodiment, the present application embodiment also provides a landslide susceptibility prediction system, such as Figure 5 As shown, Figure 5 This is a structural diagram of a landslide susceptibility prediction system provided in an embodiment of the present application. The landslide susceptibility prediction system 5 includes: an acquisition unit 501, a division unit 502, a determination unit 503 and a prediction unit 504, wherein:

[0152] The acquisition unit 501 is used to acquire a terrain image to be divided; wherein the terrain image is a slope aspect and hill shadow map extracted based on a digital terrain image;

[0153] The division unit 502 is used to divide the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units;

[0154] The determining unit 503 is configured to extract environmental factors based on the slope unit and determine a plurality of evaluation factors;

[0155] The prediction unit 504 is used to predict the landslide susceptibility of the slope unit based on the multiple evaluation factors through a landslide susceptibility prediction model to obtain a landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to predict landslides on slope units.

[0156] In some embodiments of the present application, the division unit 502 is further configured to perform region division on the terrain image using the multi-scale algorithm to obtain a plurality of image regions;

[0157] The determining unit 503 is further configured to merge any two adjacent image regions among the multiple image regions based on the principle of minimum heterogeneity to determine an initial slope unit; and to segment, merge, and smooth the initial slope unit to determine the slope unit.

[0158] In some embodiments of the present application, the acquisition unit 501 is further configured to merge any two adjacent image regions among the multiple image regions based on the minimum heterogeneity principle to obtain a target image region; wherein the area of ​​the target image region is larger than the area of ​​the image region;

[0159] The determination unit 503 is further configured to calculate the heterogeneity of the target image region; if the heterogeneity is greater than a scale threshold, the any two adjacent image regions are not merged, and the any two adjacent image regions are respectively determined as the initial slope units; if the heterogeneity is less than a scale threshold, the merging process is continued based on the target image region until the heterogeneity of the obtained new target image region is greater than the scale threshold, and the initial slope unit is determined.

[0160] In some embodiments of the present application, the acquisition unit 501 is further configured to, before performing region division on the terrain image using a multi-scale algorithm to obtain slope units, acquire historical landslide data of a study area corresponding to the terrain image to be divided;

[0161] The determination unit 503 is further used to calculate the landslide area and shape characteristics based on the historical landslide data; based on the landslide area and the shape characteristics, perform multiple parameter adjustments to determine the optimal parameters of the multi-scale algorithm; wherein the optimal parameters include scale, shape feature weight and compactness weight.

[0162] In some embodiments of the present application, the determining unit 503 is further configured to perform initial susceptibility differentiation based on the multiple evaluation factors using the self-organizing map neural network module to determine an initial susceptibility evaluation result; and to perform grading and information quantity calculation on the multiple evaluation factors using the information quantity module to determine the information quantity value corresponding to each of the multiple evaluation factors.

[0163] The prediction unit 504 is further configured to perform landslide susceptibility prediction on the slope unit based on the information amount values ​​corresponding to the multiple evaluation factors through the support vector machine module, and obtain a landslide susceptibility prediction result corresponding to the slope unit.

[0164] In some embodiments of the present application, the acquisition unit 501 is further configured to classify each of the multiple evaluation factors using the information volume module to obtain a secondary classification result for each evaluation factor;

[0165] The determination unit 503 is also used to calculate the information value of each evaluation factor based on the secondary classification result of each evaluation factor and the probability of geological disasters occurring in the study area corresponding to the acquired terrain image; and determine the information value corresponding to each of the multiple evaluation factors based on the information value of each evaluation factor.

[0166] In some embodiments of the present application, the acquisition unit 501 is further configured to obtain landslide sample data of the study area before performing landslide susceptibility prediction on the slope unit based on the multiple evaluation factors using the landslide susceptibility prediction model and obtaining the landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide sample data is the slope unit where the historical landslide occurred;

[0167] The determination unit 503 is further configured to determine a plurality of environmental evaluation factors based on the study area; perform initial susceptibility differentiation and non-landslide sample production through a self-organizing map neural network module based on the plurality of environmental evaluation factors, and determine non-landslide sample data; determine training sample data and test sample data based on the landslide sample data and the non-landslide sample data; train an initial landslide susceptibility prediction model through the training sample data to obtain a trained landslide susceptibility prediction model; test the trained landslide susceptibility prediction model through the test sample data to obtain a test result; and determine the landslide susceptibility prediction model when the test result is greater than a preset error threshold.

[0168] Based on the landslide susceptibility prediction method of the above embodiment, the embodiment of the present application also provides a landslide susceptibility prediction device, such as Figure 6 As shown, Figure 6 This is a schematic diagram of the structure of a landslide susceptibility prediction device provided in an embodiment of the present application. The landslide susceptibility prediction device 6 includes a processor 601 and a memory 602. The memory 602 is used to store a computer program; the processor 601 is used to load and execute the computer program from the memory to perform the landslide susceptibility prediction method described in the above embodiment.

[0169] In the embodiment of the present application, the processor 601 may be at least one of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, and a microprocessor. It is understood that for different devices, the electronic device used to implement the above-mentioned processor function may also be other, and the embodiment of the present application does not specifically limit this.

[0170] An embodiment of the present application provides a computer-readable storage medium storing a computer program for implementing the landslide susceptibility prediction method as described in any of the above embodiments when executed by a processor.

[0171] Illustratively, the program instructions corresponding to a landslide susceptibility prediction method in this embodiment can be stored on a storage medium such as a CD, a hard disk, or a USB flash drive. When the program instructions corresponding to a landslide susceptibility prediction method in the storage medium are read or executed by an electronic device, the landslide susceptibility prediction method described in any of the above embodiments can be implemented.

[0172] In addition, the functional modules in the embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional modules.

[0173] If the integrated unit is implemented in the form of a software functional module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method of this embodiment. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0174] It should be understood that "one embodiment" or "an embodiment" or "some embodiments" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments. The above description of the various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other. For the sake of brevity, they will not be repeated here.

[0175] The modules described above as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules; they may be located in one place or distributed across multiple network units; some or all of the modules may be selected according to actual needs to achieve the purpose of this embodiment.

[0176] In addition, all functional modules in the embodiments of the present application can be integrated into one processing unit, or each module can be a separate unit, or two or more modules can be integrated into one unit; the above-mentioned integrated modules can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0177] Those skilled in the art will understand that all or part of the steps of implementing the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, and other media that can store program codes.

[0178] The methods disclosed in the several method embodiments provided in the embodiments of this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0179] The features disclosed in several product embodiments provided in the embodiments of this application can be arbitrarily combined to obtain new product embodiments without conflict.

[0180] The features disclosed in several method or device embodiments provided in the embodiments of this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0181] The above is merely an implementation of the embodiments of the present application, but the scope of protection of the embodiments of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the embodiments of the present application. Therefore, the scope of protection of the embodiments of the present application should be based on the scope of protection of the claims.

Claims

1. A landslide susceptibility prediction method, characterized in that: The method comprises: Acquire a terrain image to be divided; wherein the terrain image is a slope aspect and hill shadow map extracted based on a digital terrain image; Dividing the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units; Based on the slope unit, environmental factors are extracted to determine multiple evaluation factors; Performing a landslide susceptibility prediction on the slope unit based on the multiple evaluation factors using a landslide susceptibility prediction model to obtain a landslide susceptibility prediction result corresponding to the slope unit; The landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained with historical landslide data and used for landslide prediction of slope units.

2. The method according to claim 1, characterized in that The terrain image is divided into regions by a multi-scale algorithm to obtain slope units, including: Dividing the terrain image into regions using the multi-scale algorithm to obtain multiple image regions; Based on the minimum heterogeneity principle, any two adjacent image areas in the multiple image areas are merged to determine an initial slope unit; The initial slope units are segmented, merged and smoothed to determine the slope units.

3. The method according to claim 2, characterized in that The step of merging any two adjacent image regions among the plurality of image regions based on the minimum heterogeneity principle to determine an initial slope unit includes: Based on the minimum heterogeneity principle, any two adjacent image regions among the multiple image regions are merged to obtain a target image region; wherein the area of ​​the target image region is larger than the area of ​​the image region; calculating the heterogeneity of the target image region; If the heterogeneity is greater than the scale threshold, the two adjacent image regions are not merged, and the two adjacent image regions are respectively determined as the initial slope units; If the heterogeneity is less than the scale threshold, the merging process is continued based on the target image region until the heterogeneity of the obtained new target image region is greater than the scale threshold, and the initial slope unit is determined.

4. The method according to any one of claims 1 to 3, characterized in that Before dividing the terrain image into regions using a multi-scale algorithm to obtain slope units, the method further includes: Acquiring historical landslide data of a study area corresponding to the terrain image to be divided; Calculating landslide area and shape characteristics based on the historical landslide data; Based on the landslide area and the shape characteristics, multiple parameter adjustments are performed to determine the optimal parameters of the multi-scale algorithm; wherein the optimal parameters include scale, shape feature weight and compactness weight.

5. The method according to claim 1, wherein The landslide susceptibility prediction model is used to predict the landslide susceptibility of the slope unit based on the multiple evaluation factors to obtain the landslide susceptibility prediction result corresponding to the slope unit, including: Based on the multiple evaluation factors, performing initial susceptibility differentiation through the self-organizing map neural network module to determine an initial susceptibility evaluation result; By means of the information quantity module, the plurality of evaluation factors are graded and information quantity values ​​are calculated, and the information quantity values ​​corresponding to the plurality of evaluation factors are determined; The support vector machine module is used to predict the landslide susceptibility of the slope unit based on the information amount values ​​corresponding to the multiple evaluation factors, so as to obtain the landslide susceptibility prediction result corresponding to the slope unit.

6. The method according to claim 5, characterized in that The step of grading the plurality of evaluation factors and calculating information quantity values ​​by the information quantity module, and determining the information quantity values ​​corresponding to the plurality of evaluation factors, includes: Using the information volume module, grading each of the multiple evaluation factors to obtain a secondary classification result of each evaluation factor; Calculating the information value of each evaluation factor based on the secondary classification result of each evaluation factor and the probability of geological disasters occurring in the study area corresponding to the acquired terrain image; Based on the information amount value of each evaluation factor, the information amount values ​​corresponding to each of the plurality of evaluation factors are determined.

7. The method according to any one of claims 1 to 3, characterized in that Before performing landslide susceptibility prediction on the slope unit based on the multiple evaluation factors using the landslide susceptibility prediction model to obtain the landslide susceptibility prediction result corresponding to the slope unit, the method further includes: Obtaining landslide sample data in the study area; wherein the landslide sample data is the slope unit where the historical landslide is located; Determining a plurality of environmental assessment factors based on the study area; Based on the multiple environmental assessment factors, performing initial susceptibility differentiation and non-landslide sample production through a self-organizing map neural network module to determine non-landslide sample data; Determining training sample data and test sample data based on the landslide sample data and the non-landslide sample data; Training the initial landslide susceptibility prediction model using the training sample data to obtain a trained landslide susceptibility prediction model; The trained landslide susceptibility prediction model is tested using the test sample data to obtain a test result; and when the test result is greater than a preset error threshold, the landslide susceptibility prediction model is determined.

8. A landslide susceptibility prediction system, characterized in that: include: Acquisition unit, division unit, determination unit and prediction unit, wherein, The acquisition unit is used to acquire a terrain image to be divided; wherein the terrain image is a slope aspect and hill shadow map extracted based on a digital terrain image; The division unit is used to divide the terrain image into regions using a multi-scale algorithm to obtain slope units; wherein the multi-scale algorithm is used to divide the terrain into slope units; The determining unit is configured to extract environmental factors based on the slope unit and determine a plurality of evaluation factors; The prediction unit is used to predict the landslide susceptibility of the slope unit based on the multiple evaluation factors through a landslide susceptibility prediction model to obtain a landslide susceptibility prediction result corresponding to the slope unit; wherein the landslide susceptibility prediction model includes: a self-organizing map neural network module, an information volume module and a support vector machine module; the landslide susceptibility prediction model is a model trained using historical landslide data and used to predict landslides on slope units.

9. A landslide susceptibility prediction device, characterized in that: include: processor and memory, wherein The memory is used to store computer programs; The processor is configured to call and run the computer program from the memory to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that Executable instructions are stored, which are used to cause a processor to execute and implement the method according to any one of claims 1 to 7.

Images