Landslide susceptibility evaluation method based on category type feature improvement

Abstract

The invention provides a landslide susceptibility evaluation method based on category-type feature improvement, and belongs to the field of landslide prediction.The method comprises the steps that obtained various types of data of a landslide disaster research area are preprocessed, evaluation factor original values corresponding to landslide points are exported, an original data set is obtained, a category-type feature improvement algorithm is trained, and a landslide susceptibility evaluation result is obtained; performing grid search to obtain a category-type feature improvement algorithm with optimal parameters, and performing adaptive enhancement integration by using the adaptive enhancement model as a meta-learner and using the category-type feature improvement algorithm with the optimal parameters as a base learner to obtain an integrated model; substituting the training set into the integrated model to perform training simulation, importing the grid points into the trained integrated model to perform landslide disaster susceptibility evaluation, and outputting a landslide disaster susceptibility prediction map; according to the method, the problems of low adaptability of existing different machine learning models to complex data and low accuracy of susceptibility evaluation are solved.

Description

Technical Field

[0001] The invention belongs to the field of landslide disaster prediction, and in particular relates to a landslide susceptibility evaluation method based on category feature enhancement. Background Art

[0002] As landslides occur frequently, causing heavy losses to people and property, accurate landslide susceptibility assessment is particularly important. Traditional machine learning methods usually have the problem of insufficient capture of nonlinear features when dealing with landslide prediction. Therefore, integrating multiple models has become an effective means to improve prediction accuracy.

[0003] A single machine learning model may encounter some limitations when dealing with complex problems, such as easy overfitting, sensitivity to outliers, and insufficient generalization ability. By combining different machine learning models, the advantages of each model can be fully utilized, such as the interpretability of decision trees and the boundary optimization ability of support vector machines. In order to overcome these shortcomings, researchers often use the integration method of multiple machine learning models, combined with search algorithms and optimization strategy frameworks to improve the performance of the model. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a landslide susceptibility assessment method based on category feature enhancement, which solves the problems of low adaptability of different existing machine learning models to complex data and low accuracy of susceptibility assessment.

[0005] In order to achieve the above objectives, the technical solution adopted by the present invention is as follows: On the one hand, the present invention provides a landslide susceptibility evaluation method based on category-based feature enhancement, comprising the following steps:

[0006] S1. Preprocess the multi-category data of the landslide hazard study area, derive the original values ​​of all initial evaluation factors corresponding to all landslide points in the landslide hazard study area, and obtain the original data set;

[0007] S2. Use the categorical feature boosting algorithm to establish a categorical feature boosting model, and use the original data set to train the categorical feature boosting model. Use the area under the curve as a training indicator, use the grid search algorithm to search for the optimal parameters of the categorical feature boosting model, and obtain the categorical feature boosting model with the optimal parameters.

[0008] S3, using the adaptive enhancement model and combining it with the category feature enhancement model with optimal parameters to perform adaptive enhancement integration, and obtain an adaptive enhancement-category feature enhancement integration model;

[0009] S4, processing the original data set to obtain an optimized data set, and using the optimized data set to perform training simulation on the adaptive enhancement-category feature enhancement integrated model to obtain a trained adaptive enhancement-category feature enhancement integrated model, and using evaluation indicators to test the accuracy of the trained adaptive enhancement-category feature enhancement integrated model;

[0010] S5. Based on the trained adaptive boosting-classification feature boosting integrated model, multiple decision trees are generated using the classification feature boosting classifier to predict the grid points in the landslide hazard study area, and the adaptive boosting classifier is used to summarize and output the susceptibility probability value corresponding to each grid unit;

[0011] S6. Use the natural breakpoint method to process the susceptibility probability value, obtain the landslide disaster susceptibility evaluation result, and output the landslide disaster susceptibility prediction map.

[0012] The beneficial effects of the present invention are as follows: the present invention obtains multi-source data of a landslide disaster research area, processes the original data, uses a software multi-value extraction to point tool, extracts the original data table of all evaluation factors of a sample set according to the landslide point data, and packages and constructs the original data set; uses a grid search algorithm to optimize the parameters of a categorical feature enhancement model, and then performs adaptive enhancement integration on the optimized categorical feature enhancement model, thereby improving the accuracy of the adaptive enhancement-categorical feature enhancement integration model; then substitutes the sample set into the adaptive enhancement-categorical feature enhancement integration model for training, thereby providing an efficient adaptive enhancement-categorical feature enhancement integration model; then imports the values ​​of all grid points into the trained model to perform landslide disaster susceptibility analysis, divides the research area into grades and outputs a landslide disaster susceptibility prediction map, thereby improving the performance index of landslide susceptibility evaluation.

[0013] Furthermore, the S1 comprises the following steps:

[0014] S101. Obtain multiple types of data on the landslide hazard study area;

[0015] S102, randomly generating non-landslide points outside a preset diameter of the landslide points according to a preset sample ratio of landslide points to non-landslide points;

[0016] S103, unifying the spatial resolution and coordinate system of the multi-category data, and converting the multi-category data after the unified coordinate system into layers of the same grid size by resampling, to obtain converted raster data;

[0017] S104, according to the converted raster data, using the software to extract to point tool, and according to the landslide point data, by extracting the original data table of all initial evaluation factors, exporting the original values ​​of all initial evaluation factors, and obtaining the original data set.

[0018] Furthermore, the S103 includes the following steps:

[0019] S1031. Unify the spatial resolution and coordinate system of multiple types of data;

[0020] S1032. Resample the multi-category data after the unified coordinate system, select the elevation grid as the reference image, set the input raster image to the raster image to be resampled, set the output raster dataset to a new raster, set the output pixel to the same elevation grid as the layer, set the resampling to the nearest, and obtain the converted raster data.

[0021] The beneficial effect of the above further scheme is that the present invention performs grid processing on the acquired multiple types of data to obtain grids with relevant settings, thereby improving the effectiveness and accuracy of the acquired data and increasing the processing speed of the enhanced model on the data.

[0022] Furthermore, the S2 comprises the following steps:

[0023] S201. Establish a categorical feature boosting model by combining the categorical feature boosting algorithm, train the categorical feature boosting model using the original data set, and perform grid search using the grid search class in the scikit-learn library;

[0024] S202. Using the fit method, the area under the curve is used as a training indicator to search for the optimal parameters of the categorical feature improvement model, obtain the optimal parameter combination, and obtain the categorical feature improvement model with the optimal parameters according to the optimal parameter combination.

[0025] Furthermore, the S201 includes the following steps:

[0026] S2011, creating a base learner including a categorical feature boosting model instance, and defining a parameter grid including a maximum number of iterations and candidate values ​​for a learning rate;

[0027] S2012. Create a grid search instance, pass in the categorical feature boosting classifier, parameter grid, cross-validation fold, scoring criteria, and parallel processing parameters, and use the grid search class to perform a grid search on the passed in parameters.

[0028] The beneficial effect of the above further scheme is that the present invention utilizes the fit method and uses the area under the curve as a training indicator, thereby improving the efficiency of the machine learning model in finding the optimal parameter combination.

[0029] Furthermore, the specific steps of S3 are as follows:

[0030] The adaptive boosting classifier in the scikit-learn library is used to call the adaptive boosting model. The adaptive boosting model is used as the meta-learner, and the categorical feature boosting model with optimal parameters is used as the base learner. The meta-learner and the base learner are adaptively enhanced and integrated to obtain the adaptive boosting-categorical feature boosting integrated model.

[0031] Furthermore, the specific steps of S4 are as follows:

[0032] S401, processing the original data set, setting the evaluation factor values ​​except ID and attribute data Y to X, setting attribute data Y to training target Y, and obtaining an optimized data set;

[0033] S402, randomly shuffling the optimized data set according to a preset set ratio and dividing it into a training set and a test set, substituting the training set into the adaptive enhancement-category feature enhancement integrated model for training simulation, and obtaining a trained adaptive enhancement-category feature enhancement integrated model;

[0034] S403. Draw a receiver operating characteristic curve using the test set to obtain the area under the curve value, and import the area under the curve function code into the adaptive enhancement-categorical feature enhancement integrated model. Use the area under the curve value to test the accuracy of the trained adaptive enhancement-categorical feature enhancement integrated model.

[0035] The beneficial effects of the above further scheme are as follows: the present invention uses the adaptive enhancement model as a meta-learner, the CatBoost algorithm as a base learner, and an integration method of multiple machine learning models, which enhances the model's adaptability to complex data and improves the predictive performance of the machine learning model.

[0036] Furthermore, the specific steps of S5 are as follows:

[0037] S501, according to the trained adaptive boosting-classification feature boosting integrated model, using the input layer to receive the evaluation factor of the study area, and using the adaptive boosting classifier to combine the base learners of multiple classification feature boosting classifiers;

[0038] S502, generating multiple decision trees using a categorical feature boosting classifier according to the optimal parameters, and predicting grid points in the landslide hazard study area according to the evaluation factors of the study area;

[0039] S503: Use the adaptive boosting classifier to summarize the prediction results of the decision tree, output the susceptibility probability value corresponding to each grid unit, and arrange the obtained susceptibility probability values ​​in ascending order.

[0040] In order to achieve the above object, according to a second aspect of the present invention, there is provided an electronic device, characterized in that it comprises a processor and a memory for storing executable instructions of the processor;

[0041] The processor is configured to execute the above-mentioned landslide susceptibility assessment method by executing.

[0042] In order to achieve the above-mentioned purpose, according to the third aspect of the present invention, a computer-readable storage medium is provided, characterized in that a plurality of classification programs are stored on the computer-readable storage medium, and the plurality of classification programs are used to be called by a processor and execute the above-mentioned landslide susceptibility assessment method.

[0043] The beneficial effect of the above further scheme is: the present invention obtains the area under the curve by drawing the receiver operating characteristic curve, and uses the area under the curve to improve the prediction performance of the machine learning model and the accuracy of the machine learning model prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a flow chart of the method.

[0045] Figure 2 Schematic diagram of the ROC curve in this embodiment.

[0046] Figure 3 Schematic diagram of performance indicators of each model in this embodiment.

[0047] Figure 4 This is a topological diagram of the adaptive enhancement-classification feature enhancement integrated model used in this embodiment to evaluate the susceptibility of landslide disasters.

[0048] Figure 5 This is a landslide disaster susceptibility prediction map in this embodiment. DETAILED DESCRIPTION

[0049] The specific implementation modes of the present invention are described below to facilitate those skilled in the art to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0050] Before describing this embodiment, the following terms are explained:

[0051] CatBoost algorithm: Category-based feature enhancement algorithm;

[0052] AdaBoost model: adaptive boost model;

[0053] CatBoostClassifier: Category-based feature boosting classifier;

[0054] AdaBoostClassifier: Adaptive boost classifier;

[0055] ROC curve: receiver operating characteristic curve;

[0056] ArcGIS: geospatial platform;

[0057] AUC: area under the curve;

[0058] NDVI: Normalized Difference Vegetation Index;

[0059] DEM: Digital Elevation Model;

[0060] TWI: Topographic Wetness Index.

[0061] Example 1

[0062] like Figure 1 As shown, the present invention provides a landslide susceptibility evaluation method based on category-based feature enhancement, and its implementation method is as follows:

[0063] S1. Preprocess the multi-category data obtained in the landslide hazard study area, derive the original values ​​of all initial evaluation factors corresponding to all landslide points in the landslide hazard study area, and obtain the original data set. The specific steps are as follows:

[0064] S101. Obtain multiple types of data on the landslide hazard study area;

[0065] S102, randomly generating non-landslide points outside a preset diameter of the landslide points according to a preset sample ratio of landslide points to non-landslide points;

[0066] S103, unifying the spatial resolution and coordinate system of the multi-class data, and converting the multi-class data after the unified coordinate system into layers of the same grid size by resampling to obtain converted raster data. The specific steps are as follows:

[0067] S1031. Unify the spatial resolution and coordinate system of multiple types of data;

[0068] S1032, resampling the multi-class data after the unified coordinate system, selecting the elevation grid as the reference image, setting the input grid image to be resampled, setting the output grid dataset to a new grid, setting the output pixel to the same elevation grid as the layer, and setting the resampling to the nearest, to obtain the converted grid data;

[0069] S104, according to the converted raster data, using the software to extract to point tool, and according to the landslide point data, by extracting the original data table of all initial evaluation factors, exporting the original values ​​of all initial evaluation factors, and obtaining the original data set.

[0070] In this embodiment, multiple types of data of the landslide disaster research area in Yuanling County, Hunan Province are obtained;

[0071] Preprocess the multi-class data of the landslide hazard study area, and randomly generate non-landslide points at a preset diameter of 1 km outside the landslide points according to the preset sample ratio of 1:1 between landslide points and non-landslide points;

[0072] Before processing the multi-class data, the coordinate system of the collected multi-class data needs to be converted to WGS_1984_UTM_Zone_46N (World Geodetic Coordinate System of 1984 - Transverse Mercator - Zone 46 of the Northern Hemisphere). Therefore, the multi-class data is imported into ArcGIS, and based on Data Management Tool—Projections and Transformations—Define Projection, the coordinate system is converted to one;

[0073] Import the data of the unified coordinate system into ArcGIS, resample the multi-category data after the unified coordinate system based on Data Management Tools—Raster—Raster Processing—Resample, select the elevation raster as the reference image, set the Input Raster to the raster image to be resampled, set the Output Raster Dataset to "newraster1", that is, the raster image after resampling, set the Output Cell Size to "Same as layer (same as layer) elevation raster", set Resample to "NEAREST", click OK, and get the converted raster data;

[0074] Based on the converted raster data and combined with the multi-value extraction to point tool in ArcGIS 10.5 software, the original data table of all initial evaluation factors was extracted according to the landslide point data, and the original numerical value export of the initial evaluation factors was completed.

[0075] S2. Use the categorical feature boosting algorithm to establish a categorical feature boosting model, and use the original data set to train the categorical feature boosting model. Use the area under the curve as a training indicator, and use the grid search algorithm to search for the optimal parameters of the categorical feature boosting model to obtain the categorical feature boosting model with the optimal parameters. The specific steps are as follows:

[0076] S201. Establish a categorical feature boosting model by combining the categorical feature boosting algorithm, train the categorical feature boosting model using the original data set, and perform grid search using the grid search class in the scikit-learn library. The specific steps are as follows:

[0077] S2011, creating a base learner including a categorical feature boosting model instance, and defining a parameter grid including a maximum number of iterations and candidate values ​​for a learning rate;

[0078] S2012, create a grid search instance, pass in the categorical feature boosting classifier, parameter grid, cross-validation fold, scoring criteria, and parallel processing parameters, and use the grid search class to perform a grid search on the passed in parameters;

[0079] S202. Using the fit method, the area under the curve is used as a training indicator to search for the optimal parameters of the categorical feature improvement model, obtain the optimal parameter combination, and obtain the categorical feature improvement model with the optimal parameters according to the optimal parameter combination.

[0080] In this embodiment, the CatBoost algorithm is used to build a CatBoost model, the data set is imported into the CatBoost model for training, AUC is used as the training indicator, and the grid search algorithm is used to search for the optimal parameters of the CatBoost model to obtain a CatBoost model with optimal parameters, as follows:

[0081] Use the GridSearchCV (grid search) class in the scikit-learn library to perform a grid search to optimize the parameters of the CatBoost model. Code: from sklearn.model_selection import GridSearchCV;

[0082] The grid search algorithm is used to search for the best parameters for the CatBoost model. AUC is used as the evaluation index to perform grid search on the two parameters of max_iter (maximum number of iterations) and learning_rate (learning rate) to obtain the best parameter combination; the parameter ranges of max_iter (maximum number of iterations) and learning_rate (learning rate) are (50, 600) and (0.01, 1) respectively. The best parameters after optimization using the grid search algorithm are max_iter (maximum number of iterations) and learning_rate (learning rate) are 100 and 0.1 respectively, as follows:

[0083] Create a CatBoostClassifier instance as a base learner and create a CatBoost classifier instance; code: CatBoost = CatBoostClassifier(random_state = 42);

[0084] Define a parameter grid param_grid, containing a series of candidate values ​​of max_iter and learning_rate; code: param_grid = {'max_iter': (50, 600), 'learning_rate': (0.01, 1)};

[0085] Create a GridSearchCV (grid search) instance, passing in the CatBoost classifier, parameter grid, cross-validation folds (cv=5), scoring criteria (scoring='auc'), and parallel processing parameters (n_jobs=-1); code: grid_search=GridSearchCV(param_grid=param_grid,cv=5,scoring='auc',n_jobs=-1);

[0086] Use the fit method to perform grid search on the training set and find the best parameter combination through the AUC evaluation index; code: grid_search.fit(X_train,y_train);

[0087] The best parameters found and the corresponding best algorithm score are output; Output the best parameters; code: print("Best parameters found:",grid_search.best_params_); Output the score of the best model; code: print("Best model score:",grid_search.best_score_);

[0088] A CatBoost model with optimal parameters was created using the best parameter combination, and predictions were made on the test set to calculate the AUC; code: y_pred = best_model.predict(X_test); Calculate the accuracy of the CatBoost model with optimal parameters on the test set; code: test_auc = (y_pred == y_test).mean(), print("Testset auc of best model:",test_auc).

[0089] S3. Using the adaptive enhancement model, combined with the category feature enhancement model with optimal parameters, adaptive enhancement integration is performed to obtain an adaptive enhancement-category feature enhancement integration model, as follows:

[0090] The adaptive boosting classifier in the scikit-learn library is used to call the adaptive boosting model. The adaptive boosting model is used as the meta-learner, and the categorical feature boosting model with optimal parameters is used as the base learner. The meta-learner and the base learner are adaptively enhanced and integrated to obtain the adaptive boosting-categorical feature boosting integrated model.

[0091] In this embodiment, the adaptive boosting classifier in the scikit-learn library is used to call the adaptive boosting model, and the CatBoost model with optimal parameters is adaptively enhanced and integrated. The code is from sklearn.ensembleimport AdaBoostClassifier; the expression is AdaBoostClassifie(CatBoostClassifier(iterations=100,learning_rate=0.1)).

[0092] S4. Process the original data set to obtain an optimized data set, and use the optimized data set to train and simulate the adaptive enhancement-category feature enhancement integrated model to obtain a trained adaptive enhancement-category feature enhancement integrated model, and use the evaluation index to test the accuracy of the trained adaptive enhancement-category feature enhancement integrated model. The specific steps are as follows:

[0093] S401, processing the original data set, setting the evaluation factor values ​​except ID and attribute data Y to X, setting attribute data Y to training target Y, and obtaining an optimized data set;

[0094] S402, randomly shuffling the optimized data set according to a preset set ratio and dividing it into a training set and a test set, substituting the training set into the adaptive enhancement-category feature enhancement integrated model for training simulation, and obtaining a trained adaptive enhancement-category feature enhancement integrated model;

[0095] S403. Draw a receiver operating characteristic curve using the test set to obtain the area under the curve value, and import the area under the curve function code into the adaptive enhancement-categorical feature enhancement integrated model. Use the area under the curve value to test the accuracy of the trained adaptive enhancement-categorical feature enhancement integrated model.

[0096] In this embodiment, the data set is processed based on Python software, the evaluation factor values ​​​​except ID and attribute data Y are set to X, and the attribute data Y is the training target Y (positive samples are 1, and negative samples are 0), and the optimized data set is obtained. The optimized data set is randomly shuffled according to the preset set ratio of 7:3, and divided into a training set and a test set, and the training set is substituted into the AdaBoost-CatBoost integrated model for training simulation;

[0097] Use the test set data to verify the model accuracy and draw the ROC curve to obtain the following Figure 2 The area under the curve AUC value shown in the figure shows that the AdaBoost ensemble model with CatBoost as the base learner can significantly increase the AUC value, and the AUC reaches 0.8970, which is better than CatBoost's 0.7930 and Adaboost's 0.6768. At the same time, the performance tests of the four indicators of AUC, ACC, Precision and F1 were also carried out, and the results are as follows Figure 3 The schematic diagram of the four performance indicators shown in the figure finally shows that the performance of the Adaboost-CatBoost integrated model is the best in each performance indicator.

[0098] S5. Based on the trained adaptive enhancement-classification feature boosting integrated model, multiple decision trees are generated using the classification feature boosting classifier to predict the grid points in the landslide hazard study area, and the adaptive enhancement classifier is used to summarize and output the susceptibility probability value corresponding to each grid unit. The specific steps are as follows:

[0099] S501, according to the trained adaptive boosting-classification feature boosting integrated model, using the input layer to receive the evaluation factor of the study area, and using the adaptive boosting classifier to combine the base learners of multiple classification feature boosting classifiers;

[0100] S502, generating multiple decision trees using a categorical feature boosting classifier according to the optimal parameters, and predicting grid points in the landslide hazard study area according to the evaluation factors of the study area;

[0101] S503: Use the adaptive boosting classifier to summarize the prediction results of the decision tree, output the susceptibility probability value corresponding to each grid unit, and arrange the obtained susceptibility probability values ​​in ascending order.

[0102] In this embodiment, Figure 4 As shown, the input layer: the adaptive enhancement-classification feature enhancement integrated model receives 16 evaluation factor information as input to the input layer; each evaluation factor is used to evaluate the susceptibility of landslides, including data on geology, topography, climate, and vegetation;

[0103] Adaptive boosting classifier: The evaluation factor received by the input layer is input into the adaptive boosting classifier. The adaptive boosting classifier is an ensemble learning method that improves the performance of the overall model by combining the prediction results of multiple categorical feature boosting classifiers as base learners;

[0104] Categorical feature boosting classifier: The adaptive boosting classifier uses the categorical feature boosting classifier as the base learner. The categorical feature boosting classifier is an algorithm based on the gradient boosting decision tree, which is configured as the best parameter combination of 100 iterations and a learning rate of 0.1 in the adaptive boosting-categorical feature boosting ensemble model;

[0105] The categorical feature boosting classifier generates multiple decision trees, each of which is a base learner. They are trained independently on the input data and output prediction results.

[0106] Decision tree generation: Multiple decision trees are generated by the categorical feature boosting classifier, including the first decision tree, the second decision tree, the third decision tree, and so on to the Nth decision tree. Each decision tree is independently trained based on the input evaluation factor information, and they together constitute the base learner set of the model;

[0107] Output layer: The prediction results of all decision trees are summarized by the adaptive boosting classifier, and the weights of each base learner are dynamically adjusted according to the performance to produce the final prediction results;

[0108] In this embodiment, the adaptive enhancement-classification feature boosting integrated model outputs two main results: AUC performance: an indicator to measure the prediction accuracy of the model. The higher the AUC value, the stronger the model's discrimination ability;

[0109] Landslide susceptibility mapping: The final output of the adaptive enhancement-categorical feature boosting ensemble model is to generate a landslide susceptibility map based on the model’s prediction results. This map can help researchers and decision makers identify areas that are more prone to landslides and take appropriate prevention and response measures in those areas.

[0110] In this example, the entire process shows the complete path from input data to final output, and how the various components in the model interact with each other to improve the accuracy of landslide susceptibility assessment. This ensemble learning method can make full use of the prediction capabilities of multiple base learners and improve the performance and robustness of the overall model by combining their prediction results.

[0111] S6. Use the natural breakpoint method to process the susceptibility probability value, obtain the landslide disaster susceptibility evaluation result, and output the landslide disaster susceptibility prediction map.

[0112] In this embodiment, the natural breakpoint method is used to divide the probability value of susceptibility into 5 levels: lower, low, medium, higher and high. The susceptibility levels are divided and judged according to the actual situation. Figure 5 The landslide susceptibility prediction map shown in Figure 2 shows that the model has a good performance in evaluating landslide susceptibility. On the one hand, the extremely low, low, medium, high and extremely high susceptibility areas of the model account for 41.0%, 15.8%, 11.7%, 11.8% and 20.1% of the total area of ​​the study area, respectively. Among the landslide geological disasters that have occurred, the landslide frequency ratio reached 6.74; on the other hand, the AUC value of the model was 0.897, indicating that the model has an excellent prediction effect.

[0113] Example 2

[0114] In this embodiment, the present invention provides a landslide susceptibility evaluation method based on category feature enhancement, which has been successfully applied to landslide disaster prediction in Yuanling County, Hunan Province. With its robustness and autonomous learning ability, it provides a new and efficient model for landslide susceptibility evaluation in Yuanling County. By optimizing the parameters of the CatBoost search algorithm and further adaptively enhancing the integration, the accuracy of the geological disaster susceptibility evaluation results is improved.

[0115] In this embodiment, this paper takes Yuanling County as the study area, integrates the local geological bureau data, the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences, and Landslides_SL to obtain a total of 289 landslides, and uses the code to randomly divide them into two groups of training data and verification data, with a preset ratio of 7:3. In this study, environmental factors were selected according to professional knowledge, and 16 environmental factors were studied: elevation, slope, aspect, surface cut coefficient, profile curvature, elevation variation coefficient, ground roughness, curvature, TWI, river distance, distance from road, distance from fault, normalized difference vegetation index, land use type, lithology and rainfall. The digital elevation model data comes from the geospatial data cloud, and the environmental factors directly obtained from the DEM are elevation, slope, aspect, surface cut coefficient, profile curvature, elevation variation coefficient, ground roughness, curvature and TWI. Among them, lithology and fault distance come from the geological cloud of the China Geological Survey, road and hydrological data, rainfall, and NDVI come from the geospatial data cloud, and land use type data comes from the data published by Wuhan University.

Claims

1. A landslide susceptibility assessment method based on category-based feature enhancement, characterized in that: The following steps are involved: S1. Preprocess the multi-category data of the landslide hazard study area, derive the original values ​​of all initial evaluation factors corresponding to all landslide points in the landslide hazard study area, and obtain the original data set; S2. Use the categorical feature boosting algorithm to establish a categorical feature boosting model, and use the original data set to train the categorical feature boosting model. Use the area under the curve as a training indicator, use the grid search algorithm to search for the optimal parameters of the categorical feature boosting model, and obtain the categorical feature boosting model with the optimal parameters. S3, using the adaptive enhancement model and combining it with the category feature enhancement model with optimal parameters to perform adaptive enhancement integration, and obtain an adaptive enhancement-category feature enhancement integration model; S4, processing the original data set to obtain an optimized data set, and using the optimized data set to perform training simulation on the adaptive enhancement-category feature enhancement integrated model to obtain a trained adaptive enhancement-category feature enhancement integrated model, and using evaluation indicators to test the accuracy of the trained adaptive enhancement-category feature enhancement integrated model; S5. Based on the trained adaptive boosting-classification feature boosting integrated model, multiple decision trees are generated using the classification feature boosting classifier to predict the grid points in the landslide hazard study area, and the adaptive boosting classifier is used to summarize and output the susceptibility probability value corresponding to each grid unit; S6. Use the natural breakpoint method to process the susceptibility probability value, obtain the landslide disaster susceptibility evaluation result, and output the landslide disaster susceptibility prediction map.

2. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 1 is characterized in that: The S1 comprises the following steps: S101. Obtain multiple types of data on the landslide hazard study area; S102, randomly generating non-landslide points outside a preset diameter of the landslide points according to a preset sample ratio of landslide points to non-landslide points; S103, unifying the spatial resolution and coordinate system of the multi-category data, and converting the multi-category data after the unified coordinate system into layers of the same grid size by resampling, to obtain converted raster data; S104, according to the converted raster data, using the software to extract to point tool, and according to the landslide point data, by extracting the original data table of all initial evaluation factors, exporting the original values ​​of all initial evaluation factors, and obtaining the original data set.

3. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 2 is characterized in that: The S103 comprises the following steps: S1031. Unify the spatial resolution and coordinate system of multiple types of data; S1032. Resample the multi-category data after the unified coordinate system, select the elevation grid as the reference image, set the input raster image to the raster image to be resampled, set the output raster dataset to a new raster, set the output pixel to the same elevation grid as the layer, set the resampling to the nearest, and obtain the converted raster data.

4. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 1 is characterized in that: The S2 comprises the following steps: S201. Establish a categorical feature boosting model by combining the categorical feature boosting algorithm, train the categorical feature boosting model using the original data set, and perform grid search using the grid search class in the scikit-learn library; S202. Using the fit method, the area under the curve is used as a training indicator to search for the optimal parameters of the categorical feature improvement model, obtain the optimal parameter combination, and obtain the categorical feature improvement model with the optimal parameters according to the optimal parameter combination.

5. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 4 is characterized in that: The S201 includes the following steps: S2011, creating a base learner including a categorical feature boosting model instance, and defining a parameter grid including a maximum number of iterations and candidate values ​​for a learning rate; S2012. Create a grid search instance, pass in the categorical feature boosting classifier, parameter grid, cross-validation fold, scoring criteria, and parallel processing parameters, and use the grid search class to perform a grid search on the passed in parameters.

6. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 1 is characterized in that: The specific steps of S3 are as follows: The adaptive boosting classifier in the scikit-learn library is used to call the adaptive boosting model. The adaptive boosting model is used as the meta-learner, and the categorical feature boosting model with optimal parameters is used as the base learner. The meta-learner and the base learner are adaptively enhanced and integrated to obtain the adaptive boosting-categorical feature boosting integrated model.

7. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 1 is characterized in that: The specific steps of S4 are as follows: S401, processing the original data set, setting the evaluation factor values ​​except ID and attribute data Y to X, setting attribute data Y to training target Y, and obtaining an optimized data set; S402, randomly shuffling the optimized data set according to a preset set ratio and dividing it into a training set and a test set, substituting the training set into the adaptive enhancement-category feature enhancement integrated model for training simulation, and obtaining a trained adaptive enhancement-category feature enhancement integrated model; S403. Draw a receiver operating characteristic curve using the test set to obtain the area under the curve value, and import the area under the curve function code into the adaptive enhancement-categorical feature enhancement integrated model. Use the area under the curve value to test the accuracy of the trained adaptive enhancement-categorical feature enhancement integrated model.

8. The landslide susceptibility assessment method based on category-based feature enhancement according to claim 1 is characterized in that: The specific steps of S5 are as follows: S501, according to the trained adaptive boosting-classification feature boosting integrated model, using the input layer to receive the evaluation factor of the study area, and using the adaptive boosting classifier to combine the base learners of multiple classification feature boosting classifiers; S502, generating multiple decision trees using a categorical feature boosting classifier according to the optimal parameters, and predicting grid points in the landslide hazard study area according to the evaluation factors of the study area; S503: Use the adaptive boosting classifier to summarize the prediction results of the decision tree, output the susceptibility probability value corresponding to each grid unit, and arrange the obtained susceptibility probability values ​​in ascending order.

9. An electronic device, characterized in that: comprising a processor and a memory for storing executable instructions of the processor; The processor is configured to implement the landslide susceptibility assessment method 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, and the plurality of classification programs are used to be called by a processor and execute the landslide susceptibility assessment method according to any one of claims 1 to 8.

Images