Integrated landslide susceptibility evaluation method combined with chimeric model
Through the Extratrees-XGB chimeric model, feature selection and weight allocation are carried out in combination with multi-source data, which solves the problems of data integration and insufficient precision in traditional landslide susceptibility assessment and realizes efficient and reliable landslide susceptibility assessment.
Patent Information
- Application Number
- CN202510856532.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-25
- Publication Date
- 2025-10-17
AI Technical Summary
Traditional landslide susceptibility assessment methods have significant shortcomings in data integration and quantitative analysis. The assessment results are not accurate enough, and the weight distribution of evaluation factors is highly subjective, making it difficult to adapt to diverse geological conditions.
The Extratrees-XGB chimeric model was used to construct a cascade model architecture through feature selection and dynamic weighting of contribution factors. Multi-source data were combined to conduct landslide susceptibility assessment. The Extratrees model was used for feature selection and preliminary modeling, and the XGB model was used to optimize decision tree construction. A weighted data set was generated for training and validation.
It improves the accuracy and robustness of landslide susceptibility assessment, realizes efficient and reliable landslide hazard assessment, and provides new technical means for the precise prevention and control of geological disasters.
Smart Images
Figure CN120804905A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of geological disaster prediction, and particularly relates to an integrated landslide susceptibility evaluation method combined with a chimeric model. BACKGROUND
[0002] The susceptibility evaluation of landslide disasters has long relied on traditional geological surveys, topographic analysis and empirical models, and these methods have significant shortcomings in data integration and quantitative analysis. Geological surveys are often limited by manpower and time costs, and it is difficult to cover large-scale complex terrain areas, while the empirical models (such as logistic regression, information content method) have limited ability to capture nonlinear relationships, resulting in insufficient accuracy of evaluation results. Especially in plateau or mountainous environments where multi-factor interactions are significant, traditional methods easily ignore potential spatial heterogeneity and cannot accurately reflect the dynamic mechanism of landslide occurrence. In addition, the weight distribution of evaluation factors in the existing evaluation process depends on expert experience, which is highly subjective and lacks data-driven optimization, further restricting the universality and reliability of the model.
[0003] With the application of machine learning technology in the field of geology, single models (such as support vector machines, random forests) have improved prediction performance to some extent, but still face key bottlenecks. For example, single models are prone to overfitting when dealing with high-dimensional, unbalanced data sets, and are insufficient in capturing complex nonlinear relationships between features, making it difficult to adapt to diverse geological conditions. At the same time, traditional feature selection methods (such as correlation analysis) fail to fully incorporate landslide occurrence mechanisms, resulting in underestimation of the contribution of important environmental factors. To address these issues, an evaluation framework that combines multi-source data and the advantages of ensemble learning is needed, which can achieve a leap from "experience-driven" to "data-mechanism dual-driven" through hierarchical feature optimization and model collaboration. SUMMARY
[0004] In view of the above problems in the prior art, the integrated landslide susceptibility evaluation method combined with a chimeric model provided by the present application solves the problems of significant shortcomings in data integration and quantitative analysis of traditional landslide susceptibility evaluation methods, insufficient accuracy of evaluation results, and strong subjectivity and lack of data-driven optimization of evaluation factors.
[0005] To achieve the above purposes, the technical scheme adopted by the present application is as follows: on the one hand, the present application provides an integrated landslide susceptibility evaluation method combined with a chimeric model, comprising the following steps: S1, using Extratrees model and XGB model to construct Extratrees-XGB chimeric model; S2, acquire the grid point data of the landslide disaster research area, and generate non-landslide points, obtain the original data set by exporting the original values of each evaluation factor corresponding to the landslide points, calculate the contribution factor value of each evaluation factor, and multiply the contribution factor value by the original value of the evaluation factor to obtain the weighted data set; S3, divide the weighted data set into a training set and a validation set, train and validate the Extratrees-XGB hybrid model using the training set and the validation set, import the grid point data into the validated Extratrees-XGB hybrid model, and perform landslide susceptibility analysis to obtain a landslide disaster susceptibility prediction map.
[0006] The present application has the following advantages: the present application uses a sample equalization process based on spatial constraints, constructs a contribution factor dynamic weighting system, and designs a cascading Extratrees-XGB hybrid model architecture, which improves the feature selection and weight distribution capabilities of the susceptibility evaluation model, and improves the precision and robustness of landslide susceptibility evaluation, and has high efficiency and reliability, providing a new technical means for precise evaluation and effective prevention and control of geological disasters.
[0007] Further, the S1 comprises the following steps: S101, import a machine learning library and create a hybrid model class; S102, based on the hybrid model class, use the Extratrees model for feature selection and preliminary modeling in the fitting method of the class, and use the XGB model to optimize the decision tree construction process in the prediction probability method of the class; S103, combine the Extratrees model and the XGB model using the hybrid model class, define the hybrid model parameters containing the Extratrees model parameters and the XGB model parameters, initialize the hybrid model class, and input the hybrid model parameters to obtain the Extratrees-XGB hybrid model.
[0008] The above further scheme has the following advantages: the present application uses the Extratrees model and the XGB model to construct the Extratrees-XGB hybrid model, deeply integrates the advantages of the Extratrees and XGB models, and improves the model convergence speed and prediction accuracy.
[0009] Further, the S2 comprises the following steps: S201, acquire multiple types of data of the landslide disaster research area, unify the spatial resolution and coordinate system of the multiple types of data, and process the unified multiple types of data into layers of the same grid size through resampling to obtain grid point data; S202, based on the landslide point in the grid point data, a same number of non-landslide points as the landslide points are randomly generated outside a preset diameter, original values of each evaluation factor corresponding to the landslide points are exported, and an original data set is obtained; S203, missing values and abnormal values in the original data set are removed, and a contribution factor value calculation expression is defined based on the statistical distribution of the landslide points and the non-landslide points, and the contribution factor value calculation expression is calculated to obtain the contribution factor value of each evaluation factor; S204, the contribution factor value of each evaluation factor is multiplied by the original value of the evaluation factor to obtain contribution data, and the contribution data is integrated to obtain a weighted data set.
[0010] Further, the contribution factor value calculation expression is as follows: ; Wherein, represents the contribution factor value, represents the frequency ratio, represents the number of evaluation factors, represents the first evaluation factor.
[0011] The above further scheme has the beneficial effects that: the application designs the contribution factor value calculation expression, calculates the contribution factor value of each factor, constructs the weighted data set, highlights the evaluation factors with larger contribution to landslide occurrence, and weakens the influence of the evaluation factors with smaller contribution, so that the feature selection is more scientific.
[0012] Further, the S3 comprises the following steps: S301, the weighted data set is divided into a training set and a validation set according to a preset proportion, the training set is used to train an Extratrees-XGB hybrid model, and the validation set is used to verify the trained Extratrees-XGB hybrid model, and a verified Extratrees-XGB hybrid model is obtained; S302, all grid point data of the landslide disaster research area are imported into the verified Extratrees-XGB hybrid model, landslide susceptibility analysis is performed, and a landslide disaster susceptibility prediction map is obtained.
[0013] Further, the S301 comprises the following steps: S3011, according to the weighted data set, the evaluation factor data except the identity number and the attribute label are set as a preset label, and the evaluation factor data containing the attribute label are set as a training target, and a set weighted data set is obtained; S3012, according to the set weighted data set, the evaluation data set is divided into a training set and a validation set based on a preset proportion by using a random sampling method; S3013, training the Extratrees-XGB hybrid model by using the training set, to obtain a trained Extratrees-XGB hybrid model; S3014, verifying the trained Extratrees-XGB hybrid model by using the verification set, and obtaining a verified Extratrees-XGB hybrid model by calculating multiple evaluation indexes.
[0014] The further scheme has the beneficial effects that the present application optimizes the division of the training set and the verification set, the parameter adjustment, and the performance evaluation, adopts a multi-index evaluation system, comprehensively quantifies the prediction ability of the model, obtains the optimal Extratrees-XGB hybrid model, and improves the prediction efficiency and accuracy.
[0015] Further, the S302 includes the following steps: S3021, inputting the original values of the evaluation factors in all grid point data in the landslide disaster research area into the verified Extratrees-XGB hybrid model, performing landslide susceptibility prediction, and outputting the susceptibility probability values of each grid unit; S3022, arranging the susceptibility probability values of each grid unit in ascending order to obtain the sorted susceptibility probability values; S3023, performing grade division by using the natural breakpoint method according to the sorted susceptibility probability values, and completing integrated landslide susceptibility evaluation.
[0016] Further, the specific steps of the landslide susceptibility prediction are as follows: According to the original values of the evaluation factors in the grid point data, an Extratrees model is used to construct multiple random decision trees, and at each node of the decision tree, a random feature is selected for splitting, and the prediction results of the decision trees are integrated to obtain an initial prediction result; An XGB model is used to gradually optimize the initial prediction result by iteratively adding decision trees based on the gradient boosting framework, to obtain and output the susceptibility probability values of each grid unit.
[0017] The further scheme has the beneficial effects that the present application processes the feature selection and preliminary modeling stage by using the Extratrees model, and optimizes the initial prediction result by using the XGB model, thereby improving the accuracy and efficiency of the hybrid model prediction.
[0018] In order to achieve the above-mentioned purposes, according to the second aspect of the present application, an electronic device is provided, characterized by comprising a processor and a memory for storing executable instructions of the processor; The processor is configured to execute the integrated landslide susceptibility assessment method combining the chimeric model.
[0019] To achieve the above object, according to a third aspect of the present application, a computer readable storage medium is provided, characterized in that the computer readable storage medium stores a plurality of classification programs for being called and executed by a processor to implement the integrated landslide susceptibility assessment method combining the chimeric model.
[0020] The above further scheme has the beneficial effect that the present application realizes landslide disaster susceptibility analysis of high performance indicators by calculating the contribution factor value and constructing the chimeric model in combination with the contribution factor value, exhibits excellent robustness and autonomous learning ability, and significantly improves the accuracy of the geological disaster susceptibility assessment result. BRIEF DESCRIPTION OF DRAWINGS
[0021] Figure 1 The method flowchart of the present application.
[0022] Figure 2 The area under the curve curve diagram drawn in the present embodiment.
[0023] Figure 3 The histogram for performance index comparison in the present embodiment.
[0024] Figure 4 The susceptibility evaluation map generated in the present embodiment. DETAILED DESCRIPTION
[0025] The specific embodiments of the present application are described below to facilitate understanding of the present application by those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, any changes within the spirit and scope of the present application as defined and determined by the appended claims are obvious, and all inventions utilizing the concept of the present application are within the scope of protection.
[0026] Before the present embodiment is described, the following terms are explained: XGB: XGBoost extreme gradient boosting; Extratrees: extreme random tree algorithm; CF value: contribution factor value; fit method: fitting method, a method for adjusting the internal parameters of a model according to the provided training data, so that the model can make predictions or decisions on unknown data; predict_proba method: a prediction probability method for providing the prediction probability of each possible label; ROC curve: receiver operating characteristic curve; AUC: area under curve; ACC: accuracy; Precision: precision.
[0027] Example 1 In this embodiment, an integrated landslide susceptibility evaluation method combining Extratrees-XGB combined chimeric model is proposed. Through multi-level feature extraction and model integration, the multi-source data is fully utilized, effectively solving the problem of insufficient feature selection and weight allocation in traditional methods, and significantly improving the precision and robustness of landslide susceptibility evaluation. In practical application, the efficiency and reliability of the integrated landslide susceptibility evaluation method combining Extratrees-XGB combined chimeric model are verified, providing a new technical means for accurate evaluation and effective prevention and control of geological disasters.
[0028] As shown in Figure 1 The present application provides an integrated landslide susceptibility evaluation method combining chimeric model, and the implementation method is as follows: S1, using Extratrees model and XGB model, constructing Extratrees-XGB chimeric model, the specific steps are as follows: S101, import machine learning library, and create chimeric model class; S102, based on the chimeric model class, in the fit method of the class, using Extratrees model for feature selection and preliminary modeling, in the prediction probability method of the class, using XGB model to optimize the decision tree construction process; S103, combining Extratrees model and XGB model by using chimeric model class, defining chimeric model parameters containing Extratrees model parameters and XGB model parameters, initializing chimeric model class, and inputting chimeric model parameters, obtaining Extratrees-XGB chimeric model.
[0029] In this embodiment, in order to build an Extratrees-XGBoost chimeric model, it is necessary to import various libraries required for machine learning, mainly including ExtratreesClassifier, XGBClassifier, dataset partitioning tool, area under the receiver operating characteristic curve calculation tool, accuracy calculation tool, F1 value calculation tool, precision calculation tool, etc.; import basic estimator and classifier hybrid classes from the sklearn.base module; import data checking functions from the sklearn.utils.validation module to verify the validity of input data; import model evaluation related functions from the sklearn.metrics module, such as area under the receiver operating characteristic curve calculation function, accuracy calculation function, F1 value calculation function, precision calculation function, recall calculation function and classification report generation function, etc., to evaluate the performance of the model.
[0030] And create a hybrid model class HybridModel, which inherits from the base estimator BaseEstimator and the classifier hybrid class ClassifierMixin; In the fit method of the class, the Extratrees model is used for feature selection and preliminary modeling. In the predict_proba method of the class, XGBoost is used to optimize the decision tree construction process and make final predictions. The corresponding primary model parameters can be passed in. If not, they are set to an empty dictionary. The secondary model is set to the XGBoost model, and the corresponding secondary model parameters can be passed in. If not, they are set to an empty dictionary.
[0031] The Extratrees model and the XGB model are combined using the chimeric model class, and the parameters of the chimeric model are defined, including the parameters of the Extratrees model and the XGB model. The chimeric model class HybridModel is initialized, and the chimeric model parameters including the parameters of the Extratrees model and the XGB model are input to obtain the Extratrees-XGB chimeric model.
[0032] In this embodiment, the prediction formula of the XGBoost model is the weighted sum of the prediction results of multiple trees, and the expression is as follows: ; in, represents the predicted value of the XGB model, Indicates the k The prediction results of the decision tree, K represents the total number of decision trees; The prediction formula of the Extratrees model is the average or majority vote of the prediction results of all trees, and the expression is as follows: ; in, represents the predicted value of the Extratrees model, represents the number of decision trees, Indicates the t The Extratrees model determines the final prediction category by calculating the average probability of each category. Specifically, each decision tree outputs the probability of each category, and then takes the average of the probabilities of each category. The final prediction category is the category with the highest probability. And adopt the cross entropy loss function in logistic regression, the expression of the loss function is: ; in, represents the loss value, represents the number of samples in the training set, Indicates the i The true label of each sample (1 indicates a landslide point, 0 indicates a non-landslide point), The model predicts the i The probability that a sample is a landslide point. The loss function is used to measure the difference between the model's predicted probability and the true label. During the training process, the model parameters are adjusted by minimizing the loss function to improve the model's prediction performance.
[0033] S2. Obtain grid point data for the landslide hazard study area and generate non-landslide points. Obtain the original data set by deriving the original values of each evaluation factor corresponding to the landslide points, and calculate the contribution factor value of each evaluation factor. Multiply the contribution factor value with the original value of the evaluation factor to obtain a weighted data set. The specific steps are as follows: S201, obtaining multi-class data of the landslide hazard study area, unifying the spatial resolution and coordinate system of the multi-class data, and processing the unified multi-class data into layers of the same grid size by resampling to obtain grid point data; S202. Based on the landslide points in the grid point data, randomly generate the same number of non-landslide points as the landslide points outside a preset diameter, and derive the original values of the evaluation factors corresponding to the landslide points to obtain an original data set.
[0034] In this embodiment, the study selects Yuanling County as the study area, integrates data from the local geological bureau, the Resource and Environment Science and Data Center of the Chinese Academy of Sciences, and Landslides_SL, and obtains 289 landslide points. In this embodiment, 16 environmental multi-class data are selected, including elevation, slope, aspect, surface cutting coefficient, profile curvature, elevation variation coefficient, ground roughness, curvature, topographic wetness index (TWI), river distance, road distance, fault distance, normalized difference vegetation index (NDVI), land use type, lithology, and rainfall; among them, the digital elevation model (DEM) data is from the geographic spatial data cloud, and the environmental factors obtained directly from the digital elevation model (DEM) include elevation, slope, aspect, surface cutting coefficient, profile curvature, elevation variation coefficient, ground roughness, curvature, and topographic wetness index (TWI); the lithology and fault distance data are from the China Geological Survey Bureau Geological Cloud; the road and hydrological data, rainfall, and normalized difference vegetation index (NDVI) are from the geographic spatial data cloud; and the land use type data is from the data set published by Wuhan University; The spatial resolution and coordinate system of the multi-class data are unified, and the coordinate system is converted to WGS_1984_UTM_Zone_46N (World Geodetic System 1984 - Transverse Mercator - Northern Hemisphere Zone 46), the evaluation factor data is imported into the geographic information system Arcgis, and the coordinate system of the evaluation factor data is unified based on Data Management Tool - Projections and Transformations - Define Projection (Data Management Tool - Projection and Transformation - Define Projection); The unified multi-class data is processed into layers of the same grid size through resampling to obtain grid point data. Specifically, the unified evaluation factor data is input into the geographic information system Arcgis again, and the multi-class data after the unified coordinate system is resampled through DataManagement Tools—Raster—Raster Processing—Resample (data management tool—raster—raster processing—resample). The elevation grid is selected as the reference image, the input raster is set as the raster image that needs to be resampled, the output raster dataset is set as "newraster1", which is the raster image after resampling, the output cell size is set as "Same as layer (the same as the layer) elevation grid", the resample is set as "NEAREST (nearest)", and the converted grid data is obtained, that is, the layers of the same grid size are obtained, and the grid point data is obtained. According to the landslide points in the grid point data, the same number of non-landslide points are randomly generated outside the landslide points with a preset diameter of 1 kilometer, and the multi-value extraction to point tool of the geographic information system Arcgis is used to export the original values of each evaluation factor corresponding to the landslide points to obtain the original data set.
[0035] S203, remove the missing values and abnormal values in the original data set, and define the contribution factor value calculation expression based on the statistical distribution of the landslide points and the non-landslide points, and calculate the contribution factor value of each evaluation factor. S204, multiply the contribution factor value of each evaluation factor with the original value of the evaluation factor to obtain the contribution data, and integrate the contribution data to obtain the weighted data set.
[0036] In this embodiment, in order to generate the weighted data set, the excel table is used to remove the missing values and abnormal values in the original data set, so as to ensure the integrity and accuracy of the data. And define the CF value, the CF value represents the contribution degree of each evaluation factor to the occurrence of landslide, based on the statistical distribution of the landslide points and the non-landslide points, the contribution factor value calculation expression is defined as shown below. ; Wherein, represents the contribution factor value, represents the frequency ratio, which is the ratio of the occurrence frequency of a certain factor value in the landslide points and the non-landslide points, represents the number of evaluation factors, represents the evaluation factor; The contribution factor value of each evaluation factor is calculated by calculating the expression of the contribution factor value. The CF value of each evaluation factor is multiplied by its original value to obtain contribution data, and the contribution data is integrated to generate a weighted data set.
[0037] S3, the weighted data set is divided into a training set and a validation set, the training set and the validation set are used to train and verify the Extratrees-XGB hybrid model, the grid point data is imported into the verified Extratrees-XGB hybrid model, and landslide susceptibility analysis is performed to obtain a landslide disaster susceptibility prediction map, and the specific steps are as follows: S301, the weighted data set is divided into a training set and a validation set according to a preset proportion, the training set is used to train the Extratrees-XGB hybrid model, and the validation set is used to verify the trained Extratrees-XGB hybrid model to obtain a verified Extratrees-XGB hybrid model, and the specific steps are as follows: S3011, according to the weighted data set, the evaluation factor data except the identity number and the attribute label is set as a preset label, and the evaluation factor data containing the attribute label is set as a training target to obtain a set weighted data set; S3012, according to the set weighted data set, the evaluation data set is divided into a training set and a validation set based on a preset proportion by using a random sampling method; S3013, the training set is used to train the Extratrees-XGB hybrid model to obtain a trained Extratrees-XGB hybrid model; S3014, the validation set is used to verify the trained Extratrees-XGB hybrid model, and a verified Extratrees-XGB hybrid model is obtained by calculating a plurality of evaluation indexes.
[0038] In this embodiment, the weighted data set is processed based on the python software, the evaluation factor values except ID and attribute label Y are set as X, and the attribute data Y is a training target Y (positive samples are 1 and negative samples are 0) to obtain a set weighted data set; According to the set weighted data set, the evaluation data set is divided into a training set and a validation set based on a preset proportion of training set:validation set=7:3 by using a random sampling method; The training set is used to train the Extratrees-XGB hybrid model to obtain a trained Extratrees-XGB hybrid model; The validation set is used to verify the trained Extratrees-XGB hybrid model, and the evaluation index AUC is calculated, and the result is as followsFigure 2 As shown in the table, it can be found that the Extratrees-XGB hybrid model can significantly increase the AUC value, and the AUC reaches 0.916, which is better than the AUC value of the single model Extratrees of 0.871 and the AUC value of the single model XGB of 0.732; And the indicators ACC, Precision and F1 are calculated, and the results are as shown in the table Figure 3 As shown in the table, it can be verified that the performance of the Extratrees-XGB hybrid model is optimal in each performance indicator, and the verified Extratrees-XGB hybrid model is obtained.
[0039] S302, import all grid point data in the landslide disaster research area into the verified Extratrees-XGB hybrid model, and perform landslide susceptibility analysis to obtain a landslide disaster susceptibility prediction map, and the specific steps are as follows: S3021, input the original values of the evaluation factors in all grid point data in the landslide disaster research area into the verified Extratrees-XGB hybrid model, perform landslide susceptibility prediction, and output the susceptibility probability values of each grid unit; S3022, arrange the susceptibility probability values of each grid unit in ascending order to obtain the sorted susceptibility probability values; S3023, according to the sorted susceptibility probability values, use the natural breakpoint method to perform grade division, and complete the integrated landslide susceptibility assessment.
[0040] In this embodiment, the verified Extratrees-XGB hybrid model is used for landslide susceptibility prediction; specifically: Input the original values of the evaluation factors in all grid point data in the landslide disaster research area into the verified Extratrees-XGB hybrid model, and perform landslide susceptibility prediction; In the feature selection and preliminary modeling stage, the Extratrees model is used, which constructs multiple random decision trees and uses random selection of features at each node for splitting, and finally integrates the prediction results of all decision trees; in the optimization and prediction stage, the XGB model is used, which is based on the gradient boosting framework and iteratively adds decision trees to gradually optimize the prediction results of the model, and obtains the susceptibility probability values of each grid unit; Output the susceptibility probability values of each grid unit, and arrange the susceptibility probability values of each grid unit in ascending order to obtain the sorted susceptibility probability values; Use the natural breakpoint method to divide the susceptibility probability values into five levels: low, low, medium, high, and high susceptibility levels, and the low, low, medium, high, and high susceptibility levels are divided and judged according to the actual situation, and the susceptibility prediction map is obtained as shown in the table Figure 4The susceptibility evaluation chart is shown; in general, the Extratrees-XGB hybrid model has better performance in landslide susceptibility evaluation, and the AUC value of the model is 0.916, indicating that the prediction effect of the model is excellent.
[0041] Embodiment 2 The embodiment provides an electronic device, including a processor and a memory for storing executable instructions of the processor; The processor is configured to realize the integrated landslide susceptibility evaluation method combined with the hybrid model as described in embodiment 1 by executing the executable instructions.
[0042] Embodiment 3 The embodiment provides a computer readable storage medium, and a plurality of classification programs are stored on the computer readable storage medium, the plurality of classification programs are used to be called and executed by a processor, and the integrated landslide susceptibility evaluation method combined with the hybrid model as described in embodiment 1 is executed.
Claims
1. An integrated landslide susceptibility assessment method combined with a chimeric model, characterized in that: The following steps are involved: S1. Using the Extratrees model and the XGB model, an Extratrees-XGB chimeric model was constructed; S2. Obtain grid point data of the landslide hazard study area and generate non-landslide points. Derive the original values of each evaluation factor corresponding to the landslide points to obtain the original data set, calculate the contribution factor value of each evaluation factor, and multiply the contribution factor value by the original value of the evaluation factor to obtain a weighted data set. S3. Divide the weighted dataset into a training set and a validation set, use the training set and validation set to train and validate the Extratrees-XGB chimeric model, import the grid point data into the validated Extratrees-XGB chimeric model, perform landslide susceptibility analysis, and obtain a landslide susceptibility prediction map.
2. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 1, characterized in that: Said S1 comprises the following steps: S101. Import the machine learning library and create a chimeric model class; S102. Based on the chimeric model class, in the class fitting method, the Extratrees model is used to perform feature selection and preliminary modeling, and in the class prediction probability method, the XGB model is used to optimize the decision tree construction process; S103. Combine the Extratrees model and the XGB model using the chimeric model class, define chimeric model parameters including Extratrees model parameters and XGB model parameters, initialize the chimeric model class, and input the chimeric model parameters to obtain the Extratrees-XGB chimeric model.
3. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 1, characterized in that: The S2 comprises the following steps: S201, obtaining multi-class data of the landslide hazard study area, unifying the spatial resolution and coordinate system of the multi-class data, and processing the unified multi-class data into layers of the same grid size by resampling to obtain grid point data; S202, based on the landslide points in the grid point data, randomly generate the same number of non-landslide points as the landslide points outside a preset diameter, and derive the original values of the evaluation factors corresponding to the landslide points to obtain an original data set; S203, removing missing values and outliers in the original data set, and defining a contribution factor value calculation expression based on the statistical distribution of landslide points and non-landslide points to calculate the contribution factor value of each evaluation factor; S204: Multiply the contribution factor value of each evaluation factor by the original value of the evaluation factor to obtain contribution data, and integrate the contribution data to obtain a weighted data set.
4. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 3, characterized in that: The contribution factor value calculation expression is as follows: in, Represents the contribution factor value, represents the frequency ratio, represents the number of evaluation factors, Indicates the evaluation factors.
5. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 1, characterized in that: The S3 includes the following steps: S301, dividing the weighted data set into a training set and a validation set according to a preset ratio, training the Extratrees-XGB chimeric model using the training set, and validating the trained Extratrees-XGB chimeric model using the validation set to obtain a validated Extratrees-XGB chimeric model; S302: Import all grid point data in the landslide hazard study area into the verified Extratrees-XGB mosaic model to perform landslide susceptibility analysis and obtain a landslide susceptibility prediction map.
6. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 5, characterized in that: The S301 includes the following steps: S3011. According to the weighted data set, the evaluation factor data other than the identity identification number and the attribute label are set as preset labels, and the evaluation factor data including the attribute label is set as a training target, thereby obtaining a set weighted data set; S3012. Divide the evaluation data set into a training set and a validation set using a random sampling method based on a preset ratio according to the set weighted data set; S3013. Training the Extratrees-XGB chimeric model using the training set to obtain a trained Extratrees-XGB chimeric model; S3014. Validate the trained Extratrees-XGB chimeric model using the validation set, and obtain a validated Extratrees-XGB chimeric model by calculating multiple evaluation indicators.
7. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 5, characterized in that: The S302 includes the following steps: S3021. Input the original values of the evaluation factors in the grid point data of all the landslide hazard study areas into the verified Extratrees-XGB mosaic model to predict the landslide susceptibility and output the susceptibility probability value of each grid cell; S3022, arranging the susceptibility probability values of each grid cell in ascending order to obtain sorted susceptibility probability values; S3023. Based on the sorted susceptibility probability values, the natural breakpoint method is used to perform grade division and complete the integrated landslide susceptibility assessment.
8. The integrated landslide susceptibility assessment method combined with a chimeric model according to claim 7, characterized in that: The specific steps of carrying out landslide susceptibility prediction are as follows: According to the original value of the evaluation factor in each grid point data, the Extratrees model is used to construct multiple random decision trees. At each node of the decision tree, randomly selected features are used for splitting, and the prediction results of the decision tree are integrated to obtain the initial prediction results. Using the XGB model and based on the gradient boosting framework, the initial prediction results are gradually optimized by iteratively adding decision trees to obtain the susceptibility probability value of each grid cell and output it.
9. An electronic device, characterized in that: comprising a processor and a memory for storing executable instructions for the processor; The processor is configured to implement the integrated landslide susceptibility assessment method combined with a chimeric model according to any one of claims 1 to 8 by executing executable instructions.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a plurality of classification programs, which are used to be called by a processor and execute the integrated landslide susceptibility assessment method combined with a chimeric model according to any one of claims 1 to 8.