Landslide prediction method and device based on data driving and electronic equipment

Abstract

The invention provides a landslide prediction method and device based on data driving and electronic equipment, and the method comprises the steps: obtaining multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; performing correlation screening on the multiple types of historical monitoring data to obtain a historical monitoring data set; obtaining a single landslide data set, a similar landslide data set and a mixed landslide data set according to the historical monitoring data set; training to obtain a plurality of landslide prediction models by adopting the single landslide data set, the similar landslide data set and the mixed landslide data set; and according to the model evaluation index, selecting a target landslide prediction model from the plurality of landslide prediction models, and performing landslide prediction on the landslide point by using the target landslide prediction model. Through adoption of the method, the problem of insufficient generalization verification of the model due to the fact that the machine learning model is applied to individual landslides and the trained model is not expanded to other places in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster prevention and control, and in particular to a landslide prediction method, device and electronic equipment based on data drive. Background Art

[0002] Landslide is one of the main types of geological disasters in China, accounting for the largest proportion in number. The main means of prevention and control is the combination of group monitoring and group prevention with professional monitoring. With the development of technologies such as the Internet of Things, big data, machine learning, and remote sensing, a large amount of geological and topographic data, landslide point monitoring data, and meteorological data have become available. Many experts and scholars have conducted qualitative and quantitative research on landslide displacement prediction using different methods, using empirical models, mathematical statistics models, nonlinear models, etc. to predict landslides. Researchers have used some advanced machine learning methods to predict landslide displacement, with high prediction accuracy, providing technical support for landslide risk assessment and early warning.

[0003] Currently, in related studies, machine learning models are usually applied to individual landslides without extending the trained models to other locations, resulting in insufficient verification of the generalizability of the models. Summary of the invention

[0004] The present invention provides a data-driven landslide prediction method, device and electronic device to solve the problem in the prior art that machine learning models are usually applied to individual landslides without extending the trained models to other locations, resulting in insufficient verification of the generalization of the models.

[0005] The present invention provides a landslide prediction method based on data drive, comprising the following steps: Obtain multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; Perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set; According to the historical monitoring data set, a single landslide data set, a similar landslide data set and a mixed landslide data set are obtained; Using single landslide dataset, similar landslide dataset and mixed landslide dataset, multiple landslide prediction models were trained. According to the model evaluation index, a target landslide prediction model is selected from the plurality of landslide prediction models, and the target landslide prediction model is used to perform landslide prediction on a landslide point.

[0006] According to the data-driven landslide prediction method provided by the present invention, a single landslide dataset, a similar landslide dataset, and a mixed landslide dataset are respectively used to train multiple landslide prediction models, including: In the training process of the plurality of landslide prediction models, the optimal hyperparameters of the landslide prediction models are selected by a grid search method.

[0007] According to the data-driven landslide prediction method provided by the present invention, a single landslide dataset, a similar landslide dataset, and a mixed landslide dataset are respectively used to train multiple landslide prediction models, including: In the training process of the plurality of landslide prediction models, the MSE indicator is used as the loss function of the landslide prediction model.

[0008] According to the data-driven landslide prediction method provided by the present invention, before performing correlation screening on multiple types of real-time monitoring data to obtain a real-time monitoring data set, the method further includes: Data cleaning is performed on multiple types of real-time monitoring data through clustering algorithm and linear difference algorithm.

[0009] According to the data-driven landslide prediction method provided by the present invention, the model evaluation index includes a mean square error and a root mean square error, and the target landslide prediction model is selected from a plurality of landslide prediction models according to the model evaluation index, including: Obtaining the mean square error and root mean square error of each of the landslide prediction models; A target landslide prediction model is selected from the plurality of landslide prediction models according to the mean square error and the root mean square error of each landslide prediction model.

[0010] According to the data-driven landslide prediction method provided by the present invention, the correlation screening of multiple types of historical monitoring data is performed to obtain a historical monitoring data set, including: Obtain the degree of relevance of any type of historical monitoring data to the landslide change stage; The correlation between any type of historical monitoring data and the landslide change stage is sorted, and the top N historical monitoring data are used as the historical monitoring data set.

[0011] The present invention also provides a data-driven landslide prediction device, comprising: A data acquisition module is used to acquire multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; A data screening module is used to screen the historical monitoring data of multiple types for relevance and obtain a historical monitoring data set; A data classification model is used to obtain a single landslide data set, a similar landslide data set, and a mixed landslide data set based on the historical monitoring data set; A model training module is used to train multiple landslide prediction models using a single landslide data set, a similar landslide data set, and a mixed landslide data set; The model application module is used to select a target landslide prediction model from a plurality of landslide prediction models according to the model evaluation index, and use the target landslide prediction model to perform landslide prediction on a landslide point.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the data-driven landslide prediction method described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the landslide prediction method based on data-driven as described above is implemented.

[0014] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the landslide prediction method based on data-driven as described above is implemented.

[0015] The data-driven landslide prediction method, device and electronic device provided by the present invention obtain a historical monitoring data set by performing correlation screening on multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; obtain a single landslide data set, a similar landslide data set and a mixed landslide data set according to the historical monitoring data set; respectively use the single landslide data set, the similar landslide data set and the mixed landslide data set to train multiple landslide prediction models; select a target landslide prediction model from multiple landslide prediction models according to model evaluation indicators, and use the target landslide prediction model to predict landslides at landslide points. This solves the problem in the related art that the machine learning model is only applied to individual landslide points, and the trained model is not extended to other locations, resulting in insufficient verification of the generalization of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is one of the flow charts of the landslide prediction method based on data-driven provided by the present invention.

[0018] Figure 2 This is the second flow chart of the data-driven landslide prediction method provided by the present invention.

[0019] Figure 3It is a structural schematic diagram of the landslide prediction device based on data drive provided by the present invention.

[0020] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the descriptions used in this way can be interchanged where appropriate, so that the embodiments can be implemented in a sequence other than that illustrated or described in the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of the steps that appear in the present application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved. The division of modules in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between units can be electrical or other similar forms, which are not limited in this application. In addition, the modules or submodules described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed in multiple circuit units, and some or all of the units may be selected according to actual needs to achieve the purpose of the present application.

[0023] Combine the following Figure 1-Figure 4 The specific contents of the present invention are described.

[0024] Figure 1It is a flow chart of a landslide prediction method based on data driving provided by the present invention, including steps S101-S105.

[0025] Step S101, obtaining multiple types of historical monitoring data uploaded by geological disaster monitoring equipment.

[0026] For example, assume that a variety of geological disaster monitoring equipment is deployed in a landslide monitoring area, such as displacement sensors, rainfall monitors, soil moisture meters, etc. These devices regularly upload data, including the displacement of the landslide body, rainfall, soil moisture, etc. We obtain historical monitoring data from these devices for the past year as subsequent training data.

[0027] In the embodiment of the present invention, the uploaded historical monitoring data of multiple types include equipment monitoring data of rainfall and surface displacement, and classification data such as slope, volume, thickness, structure type, etc. The classification data is used to determine the type of landslide occurring at the landslide point.

[0028] Step S102: performing correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set.

[0029] In an embodiment of the present invention, correlation screening is performed on multiple types of historical monitoring data to improve model accuracy and generalization ability.

[0030] In a possible implementation, Figure 2 As shown, multiple types of historical monitoring data are screened for relevance to obtain historical monitoring data sets, including: Step S201, obtaining the correlation between any type of historical monitoring data and the landslide change stage.

[0031] Step S202: sorting the correlation between any type of historical monitoring data and the landslide change stage.

[0032] Step S203: taking the top N historical monitoring data as the historical monitoring data set.

[0033] Specifically, the daily rainfall, effective rainfall, slope, volume type, thickness type, structure type, deformation movement characteristics, development stage and other features were screened, and the daily rainfall, slope, landslide volume, cracks and category codes with high correlation were selected as the input features for the next step.

[0034] Exemplarily, the XGBoost machine learning algorithm is used to perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set.

[0035] The importance of features (i.e., multiple types of historical monitoring data) is indirectly evaluated through the training results of the XGBoost model, so that important features can be screened out to form the historical monitoring data set.

[0036] For a landslide dataset containing multiple types of historical monitoring data, it includes multiple features such as daily rainfall, effective rainfall, slope, volume type, thickness type, structure type, deformation movement characteristics, development stage, etc., as well as a binary classification label indicating whether a landslide has occurred.

[0037] Perform necessary preprocessing on the data, such as missing value filling, outlier processing, data standardization, etc., to ensure data quality. If the data contains non-numerical features (such as categorical features), they need to be encoded. Divide the dataset into a training set and a validation set so that the training set can be used when training the XGBoost model and the model performance can be evaluated on the validation set.

[0038] Train the XGBoost model using the training set. During the training process, you can adjust model parameters (such as learning rate, maximum depth, minimum subtree weight, etc.) to optimize model performance.

[0039] After training is completed, the importance of each feature is evaluated using the feature importance evaluation methods provided by the XGBoost model (such as feature weight, feature gain, etc.). These evaluation methods are based on the performance of the model during training and can reflect the contribution of the feature to the model's prediction results.

[0040] According to the feature importance evaluation results, select the top N features with higher importance (such as N=5 or determined according to actual needs) as the historical monitoring data set.

[0041] For example, the first N (N=5) features are: daily rainfall in the previous four days, with a contribution of about 70%; slope, with a contribution of about 10%; landslide volume, with a contribution of about 6%; fissures, with a contribution of about 5%; and category code, with a contribution of about 4%.

[0042] In the embodiment of the present invention, XGBoost itself is not directly used for feature selection, but by analyzing the feature importance in the model training results, features that have important contributions to landslide prediction can be effectively screened out, thereby building a more accurate and efficient landslide prediction model.

[0043] In a possible implementation, before step S102, the method further includes: performing data cleaning on multiple types of real-time monitoring data by using a clustering algorithm and a linear difference algorithm.

[0044] Step S103, obtaining a single landslide data set, a similar landslide data set, and a mixed landslide data set based on the historical monitoring data set.

[0045] Specifically, in the historical monitoring data set, three training sets and validation sets are made using the data of single landslide, similar landslide and mixed landslide, and multiple landslide points are randomly selected and their data are used to make three test sets. The features with high correlation in step S102 are selected as the input for subsequent training of multiple landslide prediction models.

[0046] Step S104, using a single landslide dataset, a similar landslide dataset, and a mixed landslide dataset to train multiple landslide prediction models.

[0047] In the embodiment of the present invention, a plurality of landslide prediction models are trained by using a single landslide dataset, a similar landslide dataset and a mixed landslide dataset through LSTM algorithm.

[0048] The following is a specific description of the multiple landslide prediction models obtained by training in the embodiment of the present invention.

[0049] First, an LSTM model is constructed as a landslide prediction model. Specifically, a deep learning framework such as TensorFlow or PyTorch is used to construct the LSTM model, and parameters such as the number of LSTM layers, the number of hidden units, and the activation function of the output layer are set.

[0050] In a possible implementation, in step S104, during the training of multiple landslide prediction models, the optimal hyperparameters of the landslide prediction model are selected by a grid search method.

[0051] Specifically, define the parameter grid, including learning rate, batch size, optimizer type (such as Adam, SGD, etc.), number of LSTM layers, number of hidden units, etc. Use cross-validation to evaluate the performance of different parameter combinations on the training set. Select the parameter combination that performs best in cross-validation as the optimal hyperparameters. Train the LSTM model on the training set using the optimal hyperparameters. Monitor the loss function value (such as mean squared error MSE) during training and the performance indicators on the validation set.

[0052] In a possible implementation, in step S104, during the training of multiple landslide prediction models, the MSE indicator is used as the loss function of the landslide prediction model.

[0053] Specifically, when training the LSTM model, the mean square error (MSE) is used as the loss function. MSE measures the average of the squared errors between the model's predicted values ​​and the true values, and is suitable for regression tasks. In each training batch, the MSE between the model's predicted values ​​and the true values ​​is calculated. The back-propagation algorithm is used to update the model's weights to minimize the MSE.

[0054] Step S105, selecting a target landslide prediction model from a plurality of landslide prediction models according to the model evaluation index, and using the target landslide prediction model to perform landslide prediction on the landslide point.

[0055] In a possible implementation, the model evaluation index includes a mean square error and a root mean square error. According to the model evaluation index, a target landslide prediction model is selected from multiple landslide prediction models, including: obtaining the mean square error and the root mean square error of each landslide prediction model; and selecting the target landslide prediction model from the multiple landslide prediction models according to the mean square error and the root mean square error of each landslide prediction model.

[0056] The selected test data set was used for model prediction output. Based on the model prediction value and the actual observation value, four indicators, namely MSE (Mean Squared Error), MAE, MAPE and RMSE (Root Mean Squared Error), were statistically calculated respectively. The MSE of the same type of landslides on the test set was 8.3322, and MAPE was 0.0125; the MSE of the mixed type of landslides on the test set was 68.6586, and MAPE was 0.0623; the MSE of the single landslide model was 44.60, and MAPE was 0.044.

[0057] In an embodiment of the present invention, the smaller the mean square error and root mean square error of the landslide prediction model, the higher the selection priority of the model. Weights are set for the mean square error and root mean square error, and a target landslide prediction model is selected from multiple landslide prediction models.

[0058] In the embodiment of the present invention, the landslide types are comprehensively classified based on the landslide scale, inducing factors, kinematic characteristics, dynamic characteristics, and development stage, such as new traction residual slope accumulation landslide, new traction collapse slope accumulation landslide, revived traction collapse slope accumulation landslide, etc. The specific classification of landslides refers to the relevant technology in the field, and the embodiment of the present invention will not be described in detail here.

[0059] The present invention can achieve the following beneficial effects by adopting the above method: 1. By acquiring multiple types of historical monitoring data and training models based on single landslide, similar landslide and mixed landslide data sets, the method of the present invention can more comprehensively consider the influence of different landslide types and geological conditions, thereby improving the applicability of the model in different locations and solving the problem of insufficient model generalization verification in the prior art.

[0060] 2. The grid search method is used to select the optimal hyperparameters of the landslide prediction model, which can ensure that the model achieves optimal performance during the training process, thereby improving the accuracy of landslide prediction. It reduces the time and cost of model training; at the same time, the grid search method can significantly improve the efficiency of model training and reduce training costs.

[0061] 3. By sorting the correlation between any type of historical monitoring data and the landslide change stage, and selecting the top N data as the historical monitoring data set, the correlation between the data set and the landslide prediction can be ensured, and the composition of the data set can be optimized. The optimized historical monitoring data set is closer to the actual needs of landslide prediction, and can more effectively support the training of the landslide prediction model, thereby improving the prediction ability of the model.

[0062] The data-driven landslide prediction device provided by the present invention is described below. The data-driven landslide prediction device described below and the data-driven landslide prediction method described above can be referred to each other.

[0063] Figure 3 A schematic diagram of a data-driven landslide prediction device provided by the present invention includes the following parts: The data acquisition module 310 is used to acquire multiple types of historical monitoring data uploaded by the geological disaster monitoring equipment; A data screening module 320 is used to screen the historical monitoring data of multiple types for relevance to obtain a historical monitoring data set; A data classification model 330 is used to obtain a single landslide data set, a similar landslide data set, and a mixed landslide data set according to the historical monitoring data set; A model training module 340 is used to respectively use a single landslide data set, a similar landslide data set, and a mixed landslide data set to train multiple landslide prediction models; The model application module 350 is used to select a target landslide prediction model from the multiple landslide prediction models according to the model evaluation index, and use the target landslide prediction model to perform landslide prediction on the landslide point.

[0064] In a possible implementation manner, the model training module 340 is specifically configured to select the optimal hyperparameters of the landslide prediction model by using a grid search method during the training process of the plurality of landslide prediction models.

[0065] In a possible implementation manner, the model training module 340 is specifically configured to use the MSE indicator as the loss function of the landslide prediction model during the training process of the plurality of landslide prediction models.

[0066] In a possible implementation, the data-driven landslide prediction device further includes a data cleaning module for performing data cleaning on multiple types of real-time monitoring data by using a clustering algorithm and a linear difference algorithm.

[0067] In a possible implementation, the model application module 350 is specifically configured to obtain the mean square error and root mean square error of each of the landslide prediction models; and select a target landslide prediction model from the plurality of landslide prediction models according to the mean square error and root mean square error of each of the landslide prediction models.

[0068] In a possible implementation, the data screening module 320 is specifically used to obtain the correlation between any type of historical monitoring data and the landslide change stage; sort the correlation between any type of historical monitoring data and the landslide change stage, and use the top N historical monitoring data as the historical monitoring data set.

[0069] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communications interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute a data-driven landslide prediction method, which includes: Obtain multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; Perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set; According to the historical monitoring data set, obtain the single landslide data set, the same type of landslide data set and the mixed type of landslide data set; Using single landslide dataset, similar landslide dataset and mixed landslide dataset, multiple landslide prediction models were trained. According to the model evaluation index, a target landslide prediction model is selected from multiple landslide prediction models, and the target landslide prediction model is used to perform landslide prediction on the landslide point.

[0070] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0071] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute a data-driven landslide prediction method provided by the above methods, the method comprising: Obtain multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; Perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set; According to the historical monitoring data set, obtain the single landslide data set, the same type of landslide data set and the mixed type of landslide data set; Using single landslide dataset, similar landslide dataset and mixed landslide dataset, multiple landslide prediction models were trained. According to the model evaluation index, a target landslide prediction model is selected from multiple landslide prediction models, and the target landslide prediction model is used to perform landslide prediction on the landslide point.

[0072] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute a data-driven landslide prediction method provided by the above methods, the method comprising: Obtain multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; Perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set; According to the historical monitoring data set, obtain the single landslide data set, the same type of landslide data set and the mixed type of landslide data set; Using single landslide dataset, similar landslide dataset and mixed landslide dataset, multiple landslide prediction models were trained. According to the model evaluation index, a target landslide prediction model is selected from multiple landslide prediction models, and the target landslide prediction model is used to perform landslide prediction on the landslide point.

[0073] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0074] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data-driven landslide prediction method, characterized in that: include: Obtain multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; Perform correlation screening on multiple types of historical monitoring data to obtain a historical monitoring data set; According to the historical monitoring data set, a single landslide data set, a similar landslide data set and a mixed landslide data set are obtained; Using single landslide dataset, similar landslide dataset and mixed landslide dataset, multiple landslide prediction models were trained. According to the model evaluation index, a target landslide prediction model is selected from the plurality of landslide prediction models, and the target landslide prediction model is used to perform landslide prediction on a landslide point.

2. The data-driven landslide prediction method according to claim 1, characterized in that: The single landslide dataset, the same type of landslide dataset and the mixed type of landslide dataset are respectively used to train multiple landslide prediction models, including: In the training process of the plurality of landslide prediction models, the optimal hyperparameters of the landslide prediction models are selected by a grid search method.

3. The data-driven landslide prediction method according to claim 1, characterized in that: The single landslide dataset, the same type of landslide dataset and the mixed type of landslide dataset are respectively used to train multiple landslide prediction models, including: In the training process of the plurality of landslide prediction models, the MSE indicator is used as the loss function of the landslide prediction model.

4. The data-driven landslide prediction method according to claim 1, characterized in that: Before performing correlation screening on multiple types of real-time monitoring data to obtain a real-time monitoring data set, the following steps are also included: Data cleaning is performed on multiple types of real-time monitoring data through clustering algorithm and linear difference algorithm.

5. The data-driven landslide prediction method according to claim 1, characterized in that: The model evaluation index includes a mean square error and a root mean square error. The selecting a target landslide prediction model from a plurality of landslide prediction models according to the model evaluation index includes: Obtaining the mean square error and root mean square error of each of the landslide prediction models; A target landslide prediction model is selected from the plurality of landslide prediction models according to the mean square error and the root mean square error of each landslide prediction model.

6. The data-driven landslide prediction method according to claim 1, characterized in that: The historical monitoring data of multiple types are screened for relevance to obtain a historical monitoring data set, including: Obtain the degree of relevance of any type of historical monitoring data to the landslide change stage; The correlation between any type of historical monitoring data and the landslide change stage is sorted, and the top N historical monitoring data are used as the historical monitoring data set.

7. A data-driven landslide prediction device, characterized in that: include: A data acquisition module is used to acquire multiple types of historical monitoring data uploaded by geological disaster monitoring equipment; A data screening module is used to screen the historical monitoring data of multiple types for relevance and obtain a historical monitoring data set; A data classification model is used to obtain a single landslide data set, a similar landslide data set, and a mixed landslide data set based on the historical monitoring data set; A model training module is used to train multiple landslide prediction models using a single landslide data set, a similar landslide data set, and a mixed landslide data set; The model application module is used to select a target landslide prediction model from the plurality of landslide prediction models according to the model evaluation index, and use the target landslide prediction model to perform landslide prediction on the landslide point.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the data-driven landslide prediction method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the data-driven landslide prediction method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the data-driven landslide prediction method according to any one of claims 1 to 6 is implemented.

Images