Landslide susceptibility evaluation method and device based on superposition integration strategy

Abstract

The present application relates to the technical field of landslide disaster risk assessment, and in particular to a landslide susceptibility assessment method and device based on a superposition integration strategy, wherein the method comprises: based on a target landslide list and a target landslide disaster point distribution map, extracting at least one landslide susceptibility assessment factor within a preset spatial scale, so as to construct a training data set and a test data set to train and test multiple target basic evaluation models, so as to obtain landslide susceptibility index maps of multiple target basic evaluation models, and determine the evaluation accuracy of each target basic evaluation model; based on a preset integration strategy and evaluation accuracy, determine the final evaluation model, so as to use the final evaluation model to draw a final landslide susceptibility distinction map of the target area. The present application can integrate various basic models, maximize the performance of each type of basic model in different scenarios, ensure the accuracy of the evaluation results, and solve the problem of evaluation method selection, providing data guarantee for early warning and prevention of landslide disasters.

Description

Technical Field

[0001] The present application relates to the technical field of landslide disaster risk assessment, and in particular to a landslide susceptibility assessment method and device based on a superposition integration strategy. Background Art

[0002] Landslide susceptibility assessment reflects the spatial probability of landslide occurrence and is a key technology for landslide monitoring, early warning and assessment.

[0003] Among the related technologies, the main landslide susceptibility assessment methods are mainly divided into three categories: knowledge-based, physical model-based, and data-driven. With the breakthroughs in professional technology in remote sensing, surveying and mapping, geology and other fields, more abundant data can be obtained to complete the landslide susceptibility assessment, so the data-driven method has advantages and has higher accuracy in the evaluation process. Machine learning is a branch of the data-driven method. It can model the underlying logic only with data and predict the probability of landslides through training and learning.

[0004] However, in the relevant technologies, there is no consensus on which machine learning model is most suitable for landslide susceptibility research. Due to differences in topography, geology and other conditions in each landslide susceptibility study area, the model that obtains the best result in each study area is different. Using different machine learning models to conduct landslide susceptibility research in turn will increase the computing time and cost, and will also lead to increased complexity, but the accuracy is difficult to effectively improve, which needs to be solved urgently. Summary of the invention

[0005] The present application provides a landslide susceptibility evaluation method and device based on a superposition integration strategy to solve the problems in related technologies, such as difficulty in determining a machine learning model that can obtain the optimal results in a landslide susceptibility study area with different terrain, geological and other conditions, and the use of different machine learning models in sequence for landslide susceptibility research increases the calculation time and cost, while increasing the complexity, but the accuracy is difficult to effectively improve.

[0006] The first aspect of the present application provides a landslide susceptibility assessment method based on a superposition integration strategy, comprising the following steps: based on a target landslide list and a target landslide hazard point distribution map, extracting at least one landslide susceptibility assessment factor within a preset spatial scale, and constructing a training data set and a test data set according to the at least one landslide susceptibility assessment factor; using the training data set to train multiple target basic assessment models, so as to generate corresponding landslide susceptibility index maps using the test data set and the trained multiple target basic assessment models, so as to determine the assessment accuracy of each target basic assessment model in the multiple target basic assessment models through the landslide susceptibility index maps; determining a final assessment model from the multiple target basic assessment models based on a preset integration strategy and the evaluation accuracy, so as to use the final assessment model to draw a final landslide susceptibility distinction map of the target area, so as to obtain a landslide susceptibility assessment result of the target area.

[0007] Optionally, in one embodiment of the present application, the extracting at least one landslide susceptibility assessment factor within a preset spatial scale based on the target landslide list and the target landslide hazard point distribution map includes: determining the disaster-prone environment and inducing factors of historical landslide points based on the target landslide list and the target landslide hazard point distribution map; generating an original landslide susceptibility assessment factor in combination with the disaster-prone environment and the inducing factors to divide the original landslide susceptibility assessment factor into a plurality of continuous assessment factors and a plurality of discrete assessment factors; calculating the correlation coefficient between the plurality of continuous assessment factors to determine a plurality of valid continuous assessment factors that meet preset conditions through the correlation coefficient and the coefficient threshold; and extracting the at least one landslide susceptibility assessment factor from the plurality of discrete evaluation factors and the plurality of valid continuous evaluation factors.

[0008] Optionally, in one embodiment of the present application, the use of the test data set and the trained multiple target basic evaluation models to generate a corresponding landslide susceptibility index map to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map includes: based on the test data set and the trained target basic evaluation model, calculating the probability value of landslide occurring in each grid unit in the preset spatial scale to draw the landslide susceptibility index map according to the probability value; based on the landslide susceptibility index map, drawing the ROC curve of each target basic evaluation model in the multiple target basic evaluation models and calculating the AUC value of each target basic evaluation model in the multiple target basic evaluation models to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the ROC curve and the AUC value.

[0009] Optionally, in one embodiment of the present application, determining the final evaluation model from the multiple target basic evaluation models based on a preset integration strategy and the evaluation accuracy includes: based on the preset integration strategy, screening a target number of first-level basic evaluation models from the multiple target basic evaluation models by the evaluation accuracy; constructing a new test data set according to the landslide susceptibility index map and the at least one landslide susceptibility evaluation factor; generating a landslide susceptibility index map corresponding to the first-level basic evaluation model based on the new test data set and the first-level basic evaluation model; determining the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

[0010] Optionally, in one embodiment of the present application, the use of the final evaluation model to draw a final landslide susceptibility distinction map of the target area to obtain the landslide susceptibility evaluation result of the target area includes: using the final evaluation model to generate a final landslide susceptibility index map of the target area; combining the final landslide susceptibility index map and a preset landslide susceptibility index division strategy to draw the final landslide susceptibility distinction map to obtain the landslide susceptibility evaluation result of the target area.

[0011] The second aspect of the present application provides a landslide susceptibility assessment device based on a superposition integration strategy, including: an extraction module, which is used to extract at least one landslide susceptibility assessment factor within a preset spatial scale based on a target landslide list and a target landslide hazard point distribution map, and construct a training data set and a test data set according to the at least one landslide susceptibility assessment factor; a determination module, which is used to train multiple target basic evaluation models using the training data set, so as to generate corresponding landslide susceptibility index maps using the test data set and the trained multiple target basic evaluation models, so as to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index maps; an evaluation module, which is used to determine a final evaluation model from the multiple target basic evaluation models based on a preset integration strategy and the evaluation accuracy, so as to draw a final landslide susceptibility distinction map of a target area using the final evaluation model, so as to obtain a landslide susceptibility evaluation result of the target area.

[0012] Optionally, in one embodiment of the present application, the extraction module includes: a first determination unit, used to determine the disaster-prone environment and inducing factors of historical landslide points based on the target landslide list and the target landslide hazard point distribution map; a division unit, used to generate an original landslide susceptibility assessment factor in combination with the disaster-prone environment and the inducing factors, so as to divide the original landslide susceptibility assessment factor into multiple continuous assessment factors and multiple discrete assessment factors; a first calculation unit, used to calculate the correlation coefficient between the multiple continuous evaluation factors, so as to determine multiple valid continuous evaluation factors that meet preset conditions through the correlation coefficient and the coefficient threshold; an extraction unit, used to extract the at least one landslide susceptibility assessment factor from the multiple discrete evaluation factors and the multiple valid continuous evaluation factors.

[0013] Optionally, in one embodiment of the present application, the determination module includes: a second calculation unit, used to calculate the probability value of landslide occurring in each grid unit within the preset spatial scale based on the test data set and the trained target basic evaluation model, so as to draw the landslide susceptibility index map according to the probability value; a second determination unit, used to draw the ROC curve of each target basic evaluation model in the multiple target basic evaluation models based on the landslide susceptibility index map and calculate the AUC value of each target basic evaluation model in the multiple target basic evaluation models, so as to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the ROC curve and the AUC value.

[0014] Optionally, in one embodiment of the present application, the evaluation module includes: a screening unit, used to screen a target number of first-level basic evaluation models from the multiple target basic evaluation models through the evaluation accuracy based on the preset integration strategy; a construction unit, used to construct a new test data set according to the landslide susceptibility index map and the at least one landslide susceptibility evaluation factor; a first generation unit, used to generate a landslide susceptibility index map corresponding to the first-level basic evaluation model based on the new test data set and the first-level basic evaluation model; a third determination unit, used to determine the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

[0015] Optionally, in one embodiment of the present application, the evaluation module includes: a second generation unit, used to generate a final landslide susceptibility index map of the target area using the final evaluation model; a drawing unit, used to draw the final landslide susceptibility differentiation map in combination with the final landslide susceptibility index map and a preset landslide susceptibility index division strategy to obtain a landslide susceptibility evaluation result of the target area.

[0016] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the landslide susceptibility assessment method based on the superposition integration strategy as described in the above embodiment.

[0017] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned landslide susceptibility assessment method based on the superposition integration strategy.

[0018] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed, is used to implement the above-mentioned landslide susceptibility assessment method based on the superposition integration strategy.

[0019] The embodiment of the present application can extract the landslide susceptibility evaluation factor according to the target landslide list and the target landslide disaster point distribution map, so as to train and test multiple target basic evaluation models to obtain the final evaluation model, thereby obtaining the landslide susceptibility evaluation result of the target area through the final evaluation model. Thus, it is realized that the fusion of various basic models is realized based on the superposition integration strategy of multiple basic models, so as to maximize the performance of each basic model in different scenarios, and solve the problem of evaluation method selection while ensuring the accuracy of the final evaluation result, and the drawn landslide susceptibility zoning map also has extremely high reliability, which can provide powerful data guarantee for the early warning and prevention of landslide disasters. Thus, it solves the problem that in the related art, it is difficult to determine the machine learning model that can obtain the optimal result in the face of landslide susceptibility research areas with different conditions such as terrain and geology, and the use of different machine learning models for landslide susceptibility research in turn increases the calculation time and calculation cost, while causing the complexity to increase, but the accuracy is difficult to effectively improve.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0022] Figure 1 A flow chart of a landslide susceptibility assessment method based on a superposition integration strategy provided according to an embodiment of the present application;

[0023] Figure 2 A histogram and a density curve diagram of the elevation interval distribution of landslide hazard points according to an embodiment of the present application;

[0024] Figure 3 A schematic diagram of the ROC curve and AUC value of the basic model evaluation result and the superimposed integrated strategy evaluation result of an embodiment of the present application;

[0025] Figure 4 A flowchart of an overlay integration strategy for one embodiment of the present application;

[0026] Figure 5 A landslide susceptibility differentiation map of a superimposed integration strategy according to an embodiment of the present application;

[0027] Figure 6 This is a flow chart of a landslide susceptibility assessment method based on a superposition integration strategy according to an embodiment of the present application;

[0028] Figure 7 It is a structural schematic diagram of a landslide susceptibility assessment device based on a superposition integration strategy provided according to an embodiment of the present application;

[0029] Figure 8 It is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application.

[0030] Reference numerals:

[0031] 10-Landslide susceptibility assessment device based on superposition integration strategy: 100-extraction module, 200-determination module, 300-evaluation module; 801-memory, 802-processor, 803-communication interface. DETAILED DESCRIPTION

[0032] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0033] The following is a description of the landslide susceptibility evaluation method and device based on the superposition integration strategy of the embodiment of the present application with reference to the accompanying drawings. In view of the related art mentioned in the above background technology, it is difficult to determine the machine learning model that can obtain the best result in the face of landslide susceptibility research areas with different conditions such as terrain and geology. Using different machine learning models to study landslide susceptibility in turn increases the calculation time and calculation cost, while increasing the complexity, but the accuracy is difficult to effectively improve. The present application provides a landslide susceptibility evaluation method based on the superposition integration strategy, in which the landslide susceptibility evaluation factor can be extracted according to the target landslide list and the target landslide disaster point distribution map, so as to train and test multiple target basic evaluation models to obtain the final evaluation model, so as to obtain the landslide susceptibility evaluation result of the target area through the final evaluation model. Thus, the fusion of various basic models based on the superposition integration strategy of multiple basic models is realized, so as to maximize the performance of each basic model in different scenarios, and solve the problem of evaluation method selection while ensuring the accuracy of the final evaluation result. The drawn landslide susceptibility zoning map also has extremely high reliability, which can provide strong data guarantee for the early warning and prevention of landslide disasters. This solves the problem in related technologies that, when faced with a landslide susceptibility study area with different topographic, geological and other conditions, it is difficult to determine the machine learning model that can obtain the optimal results. Using different machine learning models to conduct landslide susceptibility studies in turn increases the computing time and cost, which leads to increased complexity, but the accuracy is difficult to effectively improve.

[0034] Specifically, Figure 1 A flow chart of a landslide susceptibility assessment method based on a superposition integration strategy provided in an embodiment of the present application.

[0035] like Figure 1 As shown in FIG. 1 , the landslide susceptibility assessment method based on the superposition integration strategy includes the following steps:

[0036] In step S101, based on the target landslide list and the target landslide hazard point distribution map, at least one landslide susceptibility evaluation factor within a preset spatial scale is extracted, and a training data set and a test data set are constructed according to the at least one landslide susceptibility evaluation factor.

[0037] It is understandable that landslide is generally defined as the rock and soil that constitutes the slope. Under the influence of gravity, due to water or other physical and chemical effects, the stress state of the slope body is changed, causing the slope body to lose stability and slide, which mainly occurs in mountainous areas. Landslide is the most common geological disaster with a high frequency of occurrence, causing more than serious casualties and economic losses every year. The landslide susceptibility assessment reflects the spatial probability of landslide occurrence and is a key technology for landslide monitoring, early warning and evaluation.

[0038] The target landslide list here can be understood as an information list based on multiple landslide hazard points and their related information in the landslide area, and the target landslide hazard point distribution map is an image drawn based on the intuitive display of the distribution and potential threats of landslide hazard points in a certain area or a specific area. The preset spatial scale here can be understood as the preset spatial range used in landslide susceptibility assessment.

[0039] Based on this, the embodiment of the present application can extract at least one landslide susceptibility evaluation factor based on a certain target landslide list and a certain landslide hazard point distribution map, so as to implement landslide susceptibility evaluation based on the landslide susceptibility evaluation factor.

[0040] For example, the present application may, but is not limited to, produce a target landslide list (including information such as landslide name, location, scale, inducing factors, historical activity content, etc.) and a target landslide hazard point distribution map through recorded historical landslide data, field investigation reports, and remote sensing image interpretation.

[0041] Based on the target landslide list and the target landslide hazard point distribution map, the embodiment of the present application can also extract at least one landslide susceptibility evaluation factor within a certain spatial scale to produce a training set and a test set.

[0042] Taking 30 meters as the spatial scale of landslide susceptibility assessment as an example, the embodiment of the present application can first extract at least one landslide susceptibility assessment factor, then resample the extracted at least one landslide susceptibility assessment factor to a spatial resolution of 30 meters, and then create a training set and a test set. The specific process can be, but is not limited to, expressed as follows:

[0043] (1) Take any number of historical landslide point data as positive sample data sets.

[0044] (2) Create a negative sample data set. The negative sample selection rule is as follows: establish a buffer zone at the landslide site where it has occurred, with a buffer zone distance of 1 km; then, extract the area outside the buffer zone of the landslide site; finally, select random point data within the area. The number of negative samples is the same as the number of positive samples.

[0045] (3) Both the negative sample dataset and the positive sample dataset are divided into training data and test data in a ratio of 8:2. The training data is used to train the models involved in the subsequent process, and the test set is used to test the accuracy of these models.

[0046] In addition, the landslide susceptibility evaluation factors in the embodiments of the present application include but are not limited to factors related to landslide development such as elevation, slope, slope aspect, curvature, engineering rock group, fault distance, fault density, hydrogeology, water system distance, water system density, annual average precipitation, average precipitation from June to September, terrain humidity index, road distance, road density and surface cover type as landslide susceptibility evaluation factors.

[0047] Optionally, in one embodiment of the present application, based on the target landslide list and the target landslide hazard point distribution map, at least one landslide susceptibility assessment factor within a preset spatial scale is extracted, including: based on the target landslide list and the target landslide hazard point distribution map, determining the disaster-prone environment and inducing factors of historical landslide points; generating original landslide susceptibility assessment factors in combination with the disaster-prone environment and inducing factors, so as to divide the original landslide susceptibility assessment factors into multiple continuous assessment factors and multiple discrete assessment factors; calculating the correlation coefficients between the multiple continuous assessment factors, so as to determine multiple valid continuous assessment factors that meet preset conditions through the correlation coefficients and the coefficient thresholds; and extracting at least one landslide susceptibility assessment factor from the multiple discrete assessment factors and the multiple valid continuous assessment factors.

[0048] In the actual implementation process, when extracting at least one landslide susceptibility evaluation factor within a certain spatial scale based on the target landslide list and the target landslide hazard point distribution map, the main purpose is but not limited to determining the disaster-prone environment and inducing factors of a historical landslide point through the two to generate an original landslide susceptibility evaluation factor, performing certain calculations and processing on the discrete evaluation factors in the original landslide susceptibility evaluation factors to obtain multiple valid continuous evaluation factors that meet preset conditions, and finally extracting at least one landslide susceptibility evaluation factor from the multiple valid continuous evaluation factors and the multiple discrete evaluation factors.

[0049] For example, the present application may first extract all original landslide susceptibility evaluation factors according to the landslide disaster-pregnant environment and inducing factors, and then process these original landslide susceptibility evaluation factors to construct a landslide susceptibility evaluation factor index system. The specific process may be, but is not limited to, expressed as follows:

[0050] (1) The evaluation factors are divided into continuous data factors and discrete data factors. For example, elevation, slope, curvature, fault distance, fault density, water system distance, water system density, annual average precipitation, average precipitation from June to September, terrain humidity index, road distance and road density are continuous evaluation factors; slope aspect, engineering rock formation, hydrogeology and land cover type are discrete data factors.

[0051] (2) Continuous data factors need to be divided into reasonable intervals for quantification and classification: In the embodiments of the present application, continuous data factors can be quantified based on, but not limited to, the distribution patterns of historical landslide disaster points. Figure 2 This is a histogram and density curve of the elevation interval distribution of landslide hazard points in one embodiment of the present application, such as Figure 2 As shown, taking elevation as an example, the present application can, but is not limited to, re-quantify the continuous evaluation factor by drawing the histogram distribution of historical landslide points in different intervals of the elevation factor and the mutation position of the density curve, that is, re-quantify the elevation factor into 5 intervals. The remaining continuous evaluation factors are graded and quantified in a similar manner.

[0052] (3) Use the correlation coefficient matrix diagnosis method as the correlation index to evaluate the correlation between continuous evaluation factors, determine the threshold, eliminate the continuous evaluation factors whose absolute value of the correlation coefficient is greater than the threshold, and use the remaining continuous evaluation factors as valid continuous evaluation factors. The calculation formula of the correlation coefficient is:

[0053]

[0054] Among them, X, Y are different evaluation factors, is the covariance between X and Y, To evaluate the variance of factor X, is the variance of the evaluation factor Y.

[0055] Taking the threshold of 0.8 as an example, in the embodiment of the present application, there are four pairs of evaluation factors whose absolute values ​​of correlation coefficients are greater than 0.8, namely, annual average precipitation and average precipitation from June to September, fault distance and fault density, road distance and road density, and water system density and water system distance. The annual average precipitation, road distance, water system distance and fault density can be retained as effective continuous evaluation factors.

[0056] Thus, the embodiment of the present application can obtain multiple effective continuous evaluation factors, combine with discrete evaluation factors, resample all evaluation factors to a certain spatial scale, extract the final landslide susceptibility evaluation factor, and construct a training set and a test set.

[0057] Step S102, training multiple target basic evaluation models using the training data set, generating corresponding landslide susceptibility index maps using the test data set and the trained multiple target basic evaluation models, and determining the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map.

[0058] As a possible implementation method, after obtaining the training data set and the test data set, the embodiment of the present application can use the training data set to train multiple target basic evaluation models, and input the test data set into these multiple trained target basic evaluation models to output their respective landslide susceptibility index maps, and evaluate the accuracy of the multiple trained target basic models based on the landslide susceptibility index map.

[0059] The target basic evaluation model here can be understood as a variety of basic models used to predict (evaluate) the probability of landslides. For example, in the embodiment of the present application, different target basic evaluation models can be established based on, but not limited to, 7 algorithms such as decision tree, logistic regression, naive Bayes, support vector machine, artificial neural network, random forest and extreme gradient boosting.

[0060] The accuracy of the evaluation results of landslide probability of each target basic evaluation model may be different when facing different terrains and geologies. Therefore, the embodiment of the present application can determine the evaluation accuracy of each target basic model through the landslide susceptibility index diagram output by each target basic evaluation model, which helps to select the most suitable basic evaluation model when facing different landslide susceptibility evaluation tasks through the evaluation accuracy.

[0061] Optionally, in one embodiment of the present application, a corresponding landslide susceptibility index map is generated using a test data set and multiple trained target basic evaluation models to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map, including: based on the test data set and the trained target basic evaluation model, calculating the probability value of a landslide occurring in each grid unit in a preset spatial scale to draw a landslide susceptibility index map according to the probability value; based on the landslide susceptibility index map, drawing the ROC curve of each target basic evaluation model in the multiple target basic evaluation models and calculating the AUC value of each target basic evaluation model in the multiple target basic evaluation models to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the ROC curve and the AUC value.

[0062] In the actual implementation process, when determining the evaluation accuracy of each target basic evaluation model, this application mainly, but not limited to, inputs the test data set into a certain trained target basic evaluation model to determine the basic model of the probability of occurrence in each grid unit at a certain spatial scale, and completes the landslide susceptibility evaluation of the basic model. The specific process can be expressed as follows:

[0063] (1) Select 7 algorithms including decision tree, logistic regression, naive Bayes, support vector machine, artificial neural network, random forest and extreme gradient boosting as the target basic model.

[0064] (2) In the Python environment, the basic evaluation models corresponding to the seven algorithms were established based on the Scikit-learn library, and the training data set and the 12 selected landslide susceptibility evaluation factors were input into each target basic evaluation model to adjust the parameters and train each target basic evaluation model. The grid search method was used to adjust the parameters with the highest degree of influence on the results in each target basic evaluation model to obtain the trained target basic evaluation model. The grid search generates all possible parameter combinations, and then trains and evaluates the model for each set of parameters, and finally selects the parameter combination with the best performance. In the embodiment of the present application, the default parameters can be selected for Naive Bayes. In logistic regression, penalty can be selected as "L2" and C can be set to 0.9; in random forest, n_estimators can be set to 180 and max_depth can be set to 10; in artificial neural network, solver can be selected as "adam"; in decision tree, max_depth can be set to 10; in support vector machine, kernel can be selected as "rbf" and C can be set to 1.5; in extreme gradient boosting, booster can be selected as "gbtree", learning_rate can be set to 0.2 and n_estimators can be set to 20.

[0065] (3) The test data set and the 12 screened landslide susceptibility evaluation factors are input into each trained target basic evaluation model, and the probability value of landslide occurrence in each grid unit of each target basic evaluation model at a certain spatial scale is calculated. This value is used as the landslide susceptibility index, and the landslide susceptibility index map of each target basic evaluation model can be drawn.

[0066] (4) Draw the operating characteristic (ROC) curve of each target basic evaluation model. By drawing the ROC curve and calculating the area under the curve (AUC), the AUC values ​​of different models can be obtained. Finally, the evaluation accuracy of each target basic evaluation model can be obtained by combining the two. Among them, the ROC curve, the full name of which is the Receiver Operating Characteristic Curve, is a graphical tool for evaluating the performance of classification models; AUC (Area Under Curve) is defined as the area under the ROC curve and the coordinate axis, and its value range is between 0.5 and 1. The larger the AUC value, the higher the prediction accuracy of the model.

[0067] Figure 3 The ROC curve and AUC value of the basic model evaluation results and the superimposed integrated strategy evaluation results of one embodiment of the present application are shown in FIG. Figure 3As shown, the decision tree AUC value is 0.865, the logistic regression AUC value is 0.909, the random forest AUC value is 0.918, the naive Bayes AUC value is 0.911, the support vector machine AUC value is 0.928, the artificial neural network AUC value is 0.925, and the extreme gradient boosting AUC value is 0.923.

[0068] It can be seen that in the landslide susceptibility evaluation results of the target basic evaluation model, except for the decision tree model, the AUC values ​​of the other basic classifiers are all higher than 0.9, among which the support vector machine (SVM) performs best with an AUC value of 0.928, followed by the artificial neural network (ANN) and extreme gradient boosting (XGboost), with AUC values ​​of 0.925 and 0.923 respectively. The evaluation accuracy of each target basic evaluation model from high to low is: support vector machine>artificial neural network>extreme gradient boosting>random forest>naive Bayes>logic regression>decision tree.

[0069] Step S103, determining a final evaluation model from multiple target basic evaluation models based on a preset integration strategy and evaluation accuracy, and using the final evaluation model to draw a final landslide susceptibility differentiation map of the target area to obtain a landslide susceptibility evaluation result of the target area.

[0070] Based on the relevant descriptions of other embodiments, it can be understood that due to the differences in topographical, geological and other conditions of each landslide study area, the model for obtaining the optimal result in each landslide study area is different. Based on this, the present application can determine the final evaluation model of the target area based on the preset integration strategy and the evaluation accuracy of each target basic evaluation model, and then use the final evaluation model to draw the final landslide susceptibility differentiation map of the target area to obtain the landslide susceptibility evaluation result of the target area.

[0071] The preset integration strategy can be understood as a certain strategy to use multiple basic evaluation models to complete the learning task. This can avoid constantly changing the basic evaluation model to determine the optimal model during the landslide probability study of the target area, effectively reducing the computational cost and time.

[0072] Optionally, in one embodiment of the present application, a final evaluation model is determined from multiple target basic evaluation models based on a preset integration strategy and evaluation accuracy, including: based on the preset integration strategy, screening a target number of first-level basic evaluation models from multiple target basic evaluation models by evaluation accuracy; constructing a new test data set based on a landslide susceptibility index map and at least one landslide susceptibility evaluation factor; based on the new test data set and the first-level basic evaluation model, generating a landslide susceptibility index map corresponding to the first-level basic evaluation model; determining the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

[0073] As a possible implementation method, when determining the final evaluation model from multiple target basic evaluation models based on a certain integration strategy and evaluation accuracy, the present application mainly but not limited to firstly screens out a target number of first-level basic evaluation models from all target basic evaluation models based on a certain integration strategy, and then constructs a new test data set through a landslide susceptibility index map and the extracted landslide susceptibility evaluation factors, and then inputs the new test data set into the first-level basic evaluation model and generates a landslide susceptibility index map, so as to determine the evaluation accuracy of each first-level basic evaluation model according to the landslide susceptibility index map of each first-level basic evaluation model, and then determines the final evaluation model according to the evaluation accuracy of each first-level basic evaluation model, and finally the final evaluation model completes the landslide susceptibility evaluation of the superimposed integration strategy.

[0074] It should be noted that the target number here refers to the screening number determined according to actual conditions, for example, three, five, etc., which can be determined or adjusted by professional and technical personnel in this field according to actual conditions. It is only used as an example here without specific limitation.

[0075] Figure 4 This is a flow chart of the stacking integration strategy of one embodiment of the present application. Figure 4 As shown, the process of determining the final evaluation model from the SOG target basic evaluation model based on a certain integration strategy and evaluation accuracy in the embodiment of the present application can be expressed as follows:

[0076] (1) Sort each target basic evaluation model based on its evaluation accuracy and select the three models with the highest evaluation accuracy, for example, support vector machine, artificial neural network, and extreme gradient boosting.

[0077] (2) The 7 landslide susceptibility index maps obtained through the 7 target basic evaluation models and the extracted landslide susceptibility evaluation factors are combined into a new feature set, which contains 19 features. The top three models in evaluation accuracy are selected as the first-level basic models, namely, the support vector machine, artificial neural network and extreme gradient boosting models. The parameters are adjusted again to obtain the landslide susceptibility index maps of the three first-level target basic evaluation models, thereby calculating the landslide susceptibility evaluation results of the three first-level target basic evaluation models, namely, the AUC values ​​of the three first-level target basic evaluation models. For example, the AUC values ​​of the support vector machine, artificial neural network and extreme gradient boosting models in this evaluation result are 0.934, 0.938 and 0.941 respectively;

[0078] (3) The newly obtained landslide susceptibility index maps (landslide susceptibility evaluation results) of the three first-level target basic evaluation models are added to the second-level feature set, forming a total of 22 features. The extreme gradient boosting model with the best performance is selected as the final evaluation model. The landslide susceptibility index map of the final evaluation model is generated by adjusting the parameters, and the landslide susceptibility evaluation results of the final evaluation model are calculated.

[0079] Furthermore, the embodiment of the present application can also evaluate the accuracy of the landslide susceptibility evaluation of the superimposed integration strategy by using the receiver operating characteristic (ROC) curve and the area under the curve (AUC), such as Figure 3 As shown, we can see that the AUC value of the superposition integration strategy is 0.95, which is higher than all basic models.

[0080] It should be noted that the number of integrations required for the application of a specific integration strategy, the number and level of target basic evaluation models selected for each integration can all be set or adjusted by professional and technical personnel in this field according to actual conditions, for example, from 9 to 7, to 5, to 3, and finally to 1, or from 9 to 5 to 1, etc. The embodiments of this application are only illustrative and not specifically limited.

[0081] The embodiment of the present application can fuse various target basic evaluation models through a superposition integration strategy, which can maximize the performance of excellent classification methods and solve the problem of evaluation method selection while ensuring accuracy.

[0082] Optionally, in one embodiment of the present application, a final landslide susceptibility differentiation map of the target area is drawn using the final evaluation model to obtain a landslide susceptibility evaluation result of the target area, including: generating a final landslide susceptibility index map of the target area using the final evaluation model; drawing a final landslide susceptibility differentiation map in combination with the final landslide susceptibility index map and a preset landslide susceptibility index division strategy to obtain a landslide susceptibility evaluation result of the target area.

[0083] In other embodiments, when the final landslide susceptibility differentiation map of the target area is drawn using the final evaluation model, the present application can generate a landslide susceptibility index map of the target area based on the final evaluation model, and then draw the final landslide susceptibility differentiation map in combination with a preset landslide susceptibility index division strategy, thereby obtaining the final evaluation result of the landslide susceptibility of the target area.

[0084] The preset landslide susceptibility index classification strategy here refers to some classification standards or rules that are preset in order to facilitate intuitive understanding of the level of landslide susceptibility.

[0085] For example, the present application may, but is not limited to, divide the final landslide susceptibility index into five intervals using 0.2, 0.4, 0.6 and 0.8 as dividing points, and define them as extremely low susceptibility area, low susceptibility area, medium susceptibility area, high susceptibility area and extremely high susceptibility area in order from low to high, to form a landslide susceptibility zoning map with a superimposed integrated model strategy. Figure 5 As shown, Figure 5 That is the landslide susceptibility zoning map of one embodiment of the present application.

[0086] The embodiment of the present application can obtain a landslide susceptibility zoning map of the target area through the final evaluation model obtained by superimposing the integration strategy, thereby providing a landslide susceptibility zoning map with extremely high reliability while ensuring the accuracy of the landslide susceptibility evaluation results, providing strong data support for the early warning and prevention of landslide disasters.

[0087] The present application is described in detail below with reference to a specific embodiment.

[0088] Figure 6 Flow chart of a landslide susceptibility assessment method based on a superposition integration strategy according to an embodiment of the present application, as shown in FIG. Figure 6 As shown:

[0089] Step S601, preparing a landslide inventory by integrating historical records, field investigation reports and remote sensing image interpretation, and drawing a landslide hazard point distribution map;

[0090] Step S602, preparing a data set of landslide susceptibility assessment factors according to the landslide list and the landslide hazard point distribution map;

[0091] Step S603, quantifying the continuous evaluation factors in the landslide susceptibility evaluation factor data, and eliminating the continuous evaluation factors with high correlation;

[0092] Step S604, constructing a training data set and a test data set;

[0093] Step S605, selecting a basic model, and using the training data set and the test data set to complete the basic model landslide susceptibility evaluation and evaluation accuracy analysis;

[0094] Step S606, drawing the landslide susceptibility evaluation and accuracy analysis of the superimposed integrated model strategy;

[0095] Step S607, drawing a landslide susceptibility classification map.

[0096] According to the landslide susceptibility evaluation method based on the superposition integration strategy proposed in the embodiment of the present application, the landslide susceptibility evaluation factor can be extracted according to the target landslide list and the target landslide disaster point distribution map, so as to train and test multiple target basic evaluation models to obtain the final evaluation model, thereby obtaining the landslide susceptibility evaluation result of the target area through the final evaluation model. Thus, the fusion of various basic models based on the superposition integration strategy of multiple basic models is realized, so as to maximize the performance of each basic model in different scenarios, solve the problem of evaluation method selection while ensuring the accuracy of the final evaluation result, and the drawn landslide susceptibility zoning map also has extremely high reliability, which can provide powerful data guarantee for the early warning and prevention of landslide disasters. Thus, the problem that it is difficult to determine the machine learning model that can obtain the optimal result in the face of landslide susceptibility research areas with different conditions such as terrain and geology in the related technology is solved, and the use of different machine learning models for landslide susceptibility research in turn increases the calculation time and calculation cost, while leading to increased complexity, but the accuracy is difficult to effectively improve.

[0097] Next, a landslide susceptibility assessment device based on a superposition integration strategy proposed in an embodiment of the present application is described with reference to the accompanying drawings.

[0098] Figure 7 It is a structural schematic diagram of a landslide susceptibility assessment device based on a superposition integration strategy according to an embodiment of the present application.

[0099] like Figure 7 As shown, the landslide susceptibility assessment device 10 based on the superposition integration strategy includes: an extraction module 100 , a determination module 200 and an assessment module 300 .

[0100] Among them, the extraction module 100 is used to extract at least one landslide susceptibility evaluation factor within a preset spatial scale based on the target landslide list and the target landslide hazard point distribution map, and construct a training data set and a test data set according to the at least one landslide susceptibility evaluation factor.

[0101] The determination module 200 is used to train multiple target basic evaluation models using the training data set, to generate corresponding landslide susceptibility index maps using the test data set and the trained multiple target basic evaluation models, and to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map.

[0102] The evaluation module 300 is used to determine a final evaluation model from multiple target basic evaluation models based on a preset integration strategy and evaluation accuracy, so as to use the final evaluation model to draw a final landslide susceptibility differentiation map of the target area to obtain a landslide susceptibility evaluation result of the target area.

[0103] Optionally, in one embodiment of the present application, the extraction module 100 includes: a first determination unit, a division unit, a first calculation unit and an extraction unit.

[0104] The first determination unit is used to determine the disaster-prone environment and inducing factors of historical landslide points based on the target landslide list and the target landslide disaster point distribution map.

[0105] The division unit is used to generate the original landslide susceptibility evaluation factor in combination with the disaster-prone environment and the inducing factors, so as to divide the original landslide susceptibility evaluation factor into a plurality of continuous evaluation factors and a plurality of discrete evaluation factors.

[0106] The first calculation unit is used to calculate the correlation coefficients between the multiple continuous evaluation factors, so as to determine multiple valid continuous evaluation factors that meet the preset conditions through the correlation coefficients and the coefficient thresholds.

[0107] The extraction unit is used to extract at least one landslide susceptibility evaluation factor from a plurality of discrete evaluation factors and a plurality of effective continuous evaluation factors.

[0108] Optionally, in one embodiment of the present application, the determination module 200 includes: a second calculation unit and a second determination unit.

[0109] Among them, the second calculation unit is used to calculate the probability value of landslide occurring in each grid unit within a preset spatial scale based on the test data set and the trained target basic evaluation model, so as to draw the landslide susceptibility index map according to the probability value.

[0110] The second determination unit is used to draw a ROC curve and calculate an AUC value according to the landslide susceptibility index diagram, so as to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the ROC curve and the AUC value.

[0111] Optionally, in one embodiment of the present application, the evaluation module includes: a screening unit, a construction unit, a generation unit and a third determination unit.

[0112] Among them, the screening unit is used to screen the first-level basic evaluation model of the target number from multiple target basic evaluation models through evaluation accuracy based on a preset integration strategy.

[0113] The construction unit is used to construct a new test data set according to the landslide susceptibility index map and at least one landslide susceptibility evaluation factor.

[0114] The generating unit is used to generate a landslide susceptibility index map corresponding to the first-level basic evaluation model based on a new test data set and the first-level basic evaluation model.

[0115] The third determination unit is used to determine the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

[0116] Optionally, in one embodiment of the present application, the evaluation module includes: a second generating unit and a drawing unit.

[0117] The second generating unit is used to generate a final landslide susceptibility index map of the target area using the final evaluation model.

[0118] The drawing unit is used to draw a final landslide susceptibility distinction map in combination with the final landslide susceptibility index map and the preset landslide susceptibility index division strategy to obtain the landslide susceptibility evaluation result of the target area.

[0119] It should be noted that the above explanations of the landslide susceptibility assessment method embodiment based on the superposition integration strategy are also applicable to the landslide susceptibility assessment device based on the superposition integration strategy of this embodiment, and will not be repeated here.

[0120] According to the landslide susceptibility evaluation device based on the superposition integration strategy proposed in the embodiment of the present application, the landslide susceptibility evaluation factor can be extracted according to the target landslide list and the target landslide disaster point distribution map, so as to train and test multiple target basic evaluation models to obtain the final evaluation model, thereby obtaining the landslide susceptibility evaluation result of the target area through the final evaluation model. Thus, the fusion of various basic models based on the superposition integration strategy of multiple basic models is realized, so as to maximize the performance of each basic model in different scenarios, solve the problem of evaluation method selection while ensuring the accuracy of the final evaluation result, and the drawn landslide susceptibility zoning map also has extremely high reliability, which can provide powerful data guarantee for the early warning and prevention of landslide disasters. Thus, the problem that it is difficult to determine the machine learning model that can obtain the optimal result in the face of landslide susceptibility research areas with different terrain, geological and other conditions in the related technology is solved, and the use of different machine learning models for landslide susceptibility research in turn increases the calculation time and calculation cost, while leading to increased complexity, but the accuracy is difficult to effectively improve.

[0121] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0122] A memory 801 , a processor 802 , and a computer program stored in the memory 801 and executable on the processor 802 .

[0123] When the processor 802 executes the program, the landslide susceptibility assessment method based on the superposition integration strategy provided in the above embodiment is implemented.

[0124] Furthermore, the electronic device further comprises:

[0125] The communication interface 803 is used for communication between the memory 801 and the processor 802 .

[0126] The memory 801 is used to store computer programs that can be executed on the processor 802 .

[0127] The memory 801 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0128] If the memory 801, the processor 802 and the communication interface 803 are implemented independently, the communication interface 803, the memory 801 and the processor 802 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0129] Optionally, in a specific implementation, if the memory 801, the processor 802 and the communication interface 803 are integrated on a chip, the memory 801, the processor 802 and the communication interface 803 can communicate with each other through an internal interface.

[0130] The processor 802 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0131] The embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned landslide susceptibility assessment method based on the superposition integration strategy.

[0132] The embodiment of the present application also provides a computer program product, including a computer program, which can run computer instructions. When the computer instructions are executed by a processor, the landslide susceptibility assessment method based on the superposition integration strategy provided in the embodiment of the present application is implemented.

[0133] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0134] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0135] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0136] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0137] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one or a combination of multiple of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0138] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0139] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0140] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A landslide susceptibility assessment method based on superposition integration strategy, characterized in that: The following steps are involved: Based on the target landslide list and the target landslide hazard point distribution map, extract at least one landslide susceptibility evaluation factor within a preset spatial scale, and construct a training data set and a test data set according to the at least one landslide susceptibility evaluation factor; Using the training data set to train multiple target basic evaluation models, using the test data set and the trained multiple target basic evaluation models to generate corresponding landslide susceptibility index maps, and determining the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map; Determining a final evaluation model from the multiple target basic evaluation models based on a preset integration strategy and the evaluation accuracy, so as to draw a final landslide susceptibility differentiation map of the target area using the final evaluation model, so as to obtain a landslide susceptibility evaluation result of the target area, wherein the preset integration strategy is a strategy of using multiple basic evaluation models in combination to complete the learning task; Among them, the determining of the final evaluation model from the multiple target basic evaluation models based on the preset integration strategy and the evaluation accuracy includes: based on the preset integration strategy, screening a target number of first-level basic evaluation models from the multiple target basic evaluation models by the evaluation accuracy; constructing a new test data set according to the landslide susceptibility index map and the at least one landslide susceptibility evaluation factor; based on the new test data set and the first-level basic evaluation model, generating a landslide susceptibility index map corresponding to the first-level basic evaluation model; determining the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

2. The landslide susceptibility assessment method based on superposition integration strategy according to claim 1 is characterized in that: The step of extracting at least one landslide susceptibility evaluation factor within a preset spatial scale based on the target landslide list and the target landslide hazard point distribution map includes: Based on the target landslide list and the distribution map of target landslide hazard sites, determine the disaster-prone environment and inducing factors of historical landslide sites; generating an original landslide susceptibility evaluation factor by combining the disaster-prone environment and the inducing factors, so as to divide the original landslide susceptibility evaluation factor into a plurality of continuous evaluation factors and a plurality of discrete evaluation factors; Calculating the correlation coefficients between the plurality of continuous evaluation factors, so as to determine a plurality of valid continuous evaluation factors satisfying a preset condition through the correlation coefficients and the coefficient threshold; The at least one landslide susceptibility evaluation factor is extracted from the plurality of discrete evaluation factors and the plurality of valid continuous evaluation factors.

3. The landslide susceptibility assessment method based on superposition integration strategy according to claim 1 is characterized in that: The step of generating a corresponding landslide susceptibility index map using the test data set and the trained multiple target basic evaluation models, so as to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map, comprises: Based on the test data set and the trained target basic evaluation model, the probability value of landslide occurring in each grid unit in the preset spatial scale is calculated to draw the landslide susceptibility index map according to the probability value; Based on the landslide susceptibility index diagram, the ROC curve of each target basic evaluation model in the multiple target basic evaluation models is drawn and the AUC value of each target basic evaluation model in the multiple target basic evaluation models is calculated, so as to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the ROC curve and the AUC value.

4. The landslide susceptibility assessment method based on superposition integration strategy according to claim 1 is characterized in that: The method of using the final evaluation model to draw a final landslide susceptibility differentiation map of the target area to obtain a landslide susceptibility evaluation result of the target area includes: generating a final landslide susceptibility index map of the target area using the final evaluation model; The final landslide susceptibility distinction map is drawn in combination with the final landslide susceptibility index map and the preset landslide susceptibility index division strategy to obtain the landslide susceptibility evaluation result of the target area.

5. A landslide susceptibility assessment device based on superposition integration strategy, characterized in that: include: An extraction module, for extracting at least one landslide susceptibility evaluation factor within a preset spatial scale based on a target landslide list and a target landslide hazard point distribution map, and constructing a training data set and a test data set according to the at least one landslide susceptibility evaluation factor; A determination module, used to train multiple target basic evaluation models using the training data set, to generate corresponding landslide susceptibility index maps using the test data set and the trained multiple target basic evaluation models, to determine the evaluation accuracy of each target basic evaluation model in the multiple target basic evaluation models through the landslide susceptibility index map; An evaluation module, for determining a final evaluation model from the multiple target basic evaluation models based on a preset integration strategy and the evaluation accuracy, so as to draw a final landslide susceptibility differentiation map of the target area using the final evaluation model, so as to obtain a landslide susceptibility evaluation result of the target area, wherein the preset integration strategy is a strategy for using multiple basic evaluation models in combination to complete the learning task; Among them, the evaluation module includes: a screening unit, used to screen a target number of first-level basic evaluation models from the multiple target basic evaluation models through the evaluation accuracy based on the preset integration strategy; a construction unit, used to construct a new test data set according to the landslide susceptibility index map and the at least one landslide susceptibility evaluation factor; a generation unit, used to generate a landslide susceptibility index map corresponding to the first-level basic evaluation model based on the new test data set and the first-level basic evaluation model; a third determination unit, used to determine the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model according to the landslide susceptibility index map corresponding to the first-level basic evaluation model, so as to determine the final evaluation model according to the evaluation accuracy of each basic evaluation model in the first-level basic evaluation model.

6. The landslide susceptibility assessment device based on superposition integration strategy according to claim 5 is characterized in that: The extraction module comprises: The first determination unit is used to determine the disaster-prone environment and inducing factors of the historical landslide points based on the target landslide list and the target landslide disaster point distribution map; A division unit, used for generating an original landslide susceptibility evaluation factor by combining the disaster-prone environment and the inducing factors, so as to divide the original landslide susceptibility evaluation factor into a plurality of continuous evaluation factors and a plurality of discrete evaluation factors; A calculation unit, used for calculating the correlation coefficients between the plurality of continuous evaluation factors, so as to determine a plurality of valid continuous evaluation factors satisfying a preset condition through the correlation coefficients and the coefficient threshold; An extraction unit is used to extract the at least one landslide susceptibility evaluation factor from the multiple discrete evaluation factors and the multiple effective continuous evaluation factors.

7. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the landslide susceptibility assessment method based on the superposition integration strategy as described in any one of claims 1 to 4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the landslide susceptibility assessment method based on the superposition integration strategy as described in any one of claims 1 to 4.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed, it is used to implement the landslide susceptibility assessment method based on the superposition integration strategy as described in any one of claims 1 to 4.

Images