Landslide displacement prediction method and system based on landslide movement mechanism and machine learning

Abstract

The invention belongs to the technical field of geological disaster prevention and control, and particularly discloses a landslide displacement prediction method and system based on a landslide movement mechanism and machine learning, and the method comprises the steps: collecting landslide displacement data, reservoir water level data and rainfall data of a research target area; selecting a fitting function according to the landslide displacement data to extract landslide motion features; performing landslide displacement correlation analysis on the reservoir water level data, the rainfall data and the landslide motion characteristics, and selecting strong correlation to construct a comprehensive data set; a CNN-LSTM machine learning model is constructed; and inputting the comprehensive data set into the CNN-LSTM machine learning model, and outputting to obtain a landslide displacement prediction result. According to the method, the problems that a single model is poor in applicability and low in precision when being used for predicting multiple types of landslide displacement-time curves are solved, and the landslide displacement can be predicted more accurately and comprehensively.

Description

Technical Field

[0001] The present invention belongs to the technical field of geological disaster prevention and control, and specifically relates to a landslide displacement prediction method and system based on landslide movement mechanism and machine learning. Background Art

[0002] According to the statistics of the past decade, landslides dominate geological disasters, accounting for 69.34% of the total geological disasters. The frequent and widespread distribution of landslides has brought far-reaching impacts and serious damage. Therefore, the research and prevention of landslides is particularly important.

[0003] Given the complexity, randomness and uncertainty of landslides, landslide prediction has always been an important and challenging topic in the field of geological disaster research. At present, landslide prediction models can be summarized into the following four categories: (1) Deterministic prediction models: This type of model mainly digitizes the internal and external influencing factors of landslides and uses mathematical and physical methods to derive accurate prediction criteria. (2) Statistical prediction models: This model is mainly based on modern mathematical theories and statistical methods, including representative models such as the Verhulst model, the gray GM (1, 1) model, the plant growth model, the exponential smoothing method and the time series prediction model. (3) Nonlinear prediction models: When dealing with the deformation and destruction evolution of landslides, this type of model introduces advanced theories such as nonlinear theory, system science theory, synergy theory, mutation theory, fractal theory, chaos theory and neural network theory for dealing with complex systems. (4) Comprehensive prediction models: This type of model mainly uses the method of multi-model fusion to predict landslides.

[0004] Although machine learning methods have obvious advantages in the field of data mining, there are still many problems and shortcomings in their application in the research of landslide displacement prediction. In the research field of landslide displacement prediction, reservoir bank landslides have attracted much attention due to their unique and complex geological conditions. It is worth noting that even in the face of the same inducing factors, different landslides will show different evolution processes and failure mechanisms. This phenomenon leads to the fact that a single landslide prediction method often has poor results in practical applications and it is difficult to accurately predict the displacement of different landslides. However, current research has not yet effectively combined the evolution mechanism of landslides with displacement prediction methods for comprehensive consideration to improve the accuracy and performance of predictions. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of poor applicability and low accuracy of a single model for predicting displacement-time curves of multiple types of landslides, and propose a landslide displacement prediction method and system based on landslide movement mechanism and machine learning.

[0006] The technical solution of the present invention is as follows: In the first aspect, a landslide displacement prediction method based on landslide movement mechanism and machine learning comprises the following steps:

[0007] S1. Collect landslide displacement data, reservoir water level data and rainfall data in the research target area;

[0008] S2. Select a fitting function based on the landslide displacement data to extract the landslide movement characteristics;

[0009] S3. Conduct landslide displacement correlation analysis on reservoir water level data, rainfall data and landslide movement characteristics, and select strong correlations to construct a comprehensive data set;

[0010] S4. Build CNN-LSTM machine learning model;

[0011] S5. Input the data in the comprehensive dataset into the CNN-LSTM machine learning model and output the landslide displacement prediction results.

[0012] The beneficial effects of the present invention are:

[0013] By selecting the corresponding fitting function for fitting, analyzing the kinematic characteristics of the landslide, combining external influencing factors such as reservoir water level data and rainfall data, and using machine learning methods for prediction, it is suitable for accurate prediction and early warning of various types of landslides, and can better provide reference suggestions for landslide disaster prevention and mitigation work.

[0014] Preferably, the step S2 specifically includes the following sub-steps:

[0015] S21. Perform data cleaning, outlier monitoring and elimination, standardization and normalization on the landslide displacement data to obtain pre-processed landslide displacement data;

[0016] S22. Draw a landslide displacement curve based on the preprocessed landslide displacement data;

[0017] S23. Analyze the dynamic characteristics of the landslide displacement curve; the dynamic characteristics include overall shape, fluctuation, mutation and trend change;

[0018] S24. According to the dynamic characteristic analysis results of the landslide displacement curve, a mathematical model is selected, and a fitting function is obtained by fitting using a nonlinear optimization algorithm;

[0019] S25. Extracting landslide movement characteristics according to the fitting function; wherein the landslide movement characteristics include the landslide deformation stage, the landslide deformation speed and the landslide deformation acceleration; the landslide deformation stage includes the initial deformation stage, the constant speed deformation stage, the accelerated deformation stage and the possible stable or broken stage.

[0020] Preferably, the step S24 specifically includes the following sub-steps:

[0021] S241. Select a mathematical model based on the dynamic characteristics analysis results of the landslide displacement curve;

[0022] S242. Initialize the parameter values ​​of the mathematical model;

[0023] S243. Iteratively updating the parameter values ​​of the mathematical model using a nonlinear optimization algorithm;

[0024] S244. Calculate the error between the predicted value obtained based on the mathematical model and the actual observed value;

[0025] S245. Determine whether the error reaches the preset convergence condition or whether the number of iterations reaches the maximum limit. If so, the process ends and the parameter values ​​of the updated mathematical model are output to obtain the fitting function; if not, return to step S243.

[0026] Preferably, in step S3, the Pearson correlation coefficient is used to perform correlation analysis;

[0027] The calculation formula of the Pearson correlation coefficient is:

[0028]

[0029] Among them, r represents the Pearson correlation coefficient, x i and i represents the values ​​of two variables at the i-th observation, and represents the mean of the two variables in the sample, and n represents the total number of observations.

[0030] Preferably, the data of the comprehensive data set in step S3 include short-term daily rainfall intensity, short-term six-day rainfall intensity, short-term seven-day rainfall intensity, long-term one-month rainfall intensity, long-term two-month rainfall intensity, daily reservoir water level fluctuations, short-term reservoir water level fluctuations, the stage of landslide deformation, the speed of landslide deformation and the acceleration of landslide deformation.

[0031] Preferably, the step S5 specifically includes the following sub-steps:

[0032] S51. Input the data in the comprehensive data set into the convolution layer to extract key features, and output the key features;

[0033] S52. Input the key features into the pooling layer for downsampling, and output the features after dimensionality reduction;

[0034] S53. Input the reduced-dimensional features into the LSTM layer to capture the long-term dependency of the landslide displacement time series, and output the landslide displacement prediction results.

[0035] The beneficial effects of the above preferred solution are:

[0036] 1. According to the dynamic characteristics analysis results of the landslide displacement curve, the corresponding mathematical model is selected, and the mathematical model is accurately fitted using a nonlinear optimization algorithm. The obtained fitting function can accurately and efficiently predict the landslide movement characteristics, which helps to improve the accuracy and performance of landslide displacement prediction.

[0037] 2. Correlation analysis using the Pearson correlation coefficient can accurately identify factors that are strongly correlated with landslide displacement, and then construct a data set that can comprehensively reflect the deformation characteristics of the landslide, which helps to improve the accuracy and performance of landslide displacement prediction.

[0038] In the second aspect, a landslide displacement prediction system based on landslide movement mechanism and machine learning includes:

[0039] Data acquisition module, used to collect landslide displacement data, reservoir water level data and rainfall data in the research target area;

[0040] A feature extraction module, used for receiving the landslide displacement data, and selecting a fitting function according to the landslide displacement data to extract the landslide movement feature;

[0041] A comprehensive data set construction module is used to receive the reservoir water level data, rainfall data and landslide movement characteristics to perform landslide displacement correlation analysis, and select strong correlation to construct a comprehensive data set;

[0042] Model building module, used to build CNN-LSTM machine learning model;

[0043] The landslide displacement prediction result module is used to receive the comprehensive data set, input the comprehensive data set into the CNN-LSTM machine learning model, and output the landslide displacement prediction result.

[0044] In a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method of the first aspect.

[0045] According to a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method according to the first aspect.

[0046] According to a fifth aspect, a computer program product is provided, comprising a computer program, wherein the computer program implements the method according to the first aspect when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Shown is a flow chart of a landslide displacement prediction method based on landslide movement mechanism and machine learning. DETAILED DESCRIPTION

[0048] Now, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the accompanying drawings are only exemplary and are intended to explain the principles and spirit of the present invention, rather than to limit the scope of the present invention.

[0049] Embodiment 1:

[0050] like Figure 1 As shown, a landslide displacement prediction method based on landslide movement mechanism and machine learning includes the following steps:

[0051] S1. Collect landslide displacement data, reservoir water level data and rainfall data in the research target area;

[0052] S2. Select a fitting function based on the landslide displacement data to extract the landslide movement characteristics;

[0053] S3. Conduct landslide displacement correlation analysis on reservoir water level data, rainfall data and landslide movement characteristics, and select strong correlations to construct a comprehensive data set;

[0054] S4. Build CNN-LSTM machine learning model;

[0055] S5. Input the data in the comprehensive dataset into the CNN-LSTM machine learning model and output the landslide displacement prediction results.

[0056] In this embodiment, step S2 specifically includes the following sub-steps:

[0057] S21. Perform data cleaning, outlier monitoring and elimination, standardization and normalization on the landslide displacement data to obtain pre-processed landslide displacement data;

[0058] S22. Draw a landslide displacement curve based on the preprocessed landslide displacement data;

[0059] S23. Analyze the dynamic characteristics of the landslide displacement curve; the dynamic characteristics include overall shape, fluctuation, mutation and trend change;

[0060] S24. According to the dynamic characteristic analysis results of the landslide displacement curve, a mathematical model is selected, and a fitting function is obtained by fitting using a nonlinear optimization algorithm;

[0061] S25. Extracting landslide movement characteristics according to the fitting function; wherein the landslide movement characteristics include the landslide deformation stage, the landslide deformation speed and the landslide deformation acceleration; the landslide deformation stage includes the initial deformation stage, the constant speed deformation stage, the accelerated deformation stage and the possible stable or broken stage.

[0062] In this embodiment, step S24 specifically includes the following sub-steps:

[0063] S241. Select a mathematical model based on the dynamic characteristics analysis results of the landslide displacement curve;

[0064] S242. Initialize the parameter values ​​of the mathematical model;

[0065] S243. Iteratively updating the parameter values ​​of the mathematical model using a nonlinear optimization algorithm;

[0066] S244. Calculate the error between the predicted value obtained based on the mathematical model and the actual observed value;

[0067] S245. Determine whether the error reaches the preset convergence condition or whether the number of iterations reaches the maximum limit. If so, the process ends and the parameter values ​​of the updated mathematical model are output to obtain the fitting function; if not, return to step S243.

[0068] In this embodiment, the Pearson correlation coefficient is used to perform correlation analysis in step S3;

[0069] The calculation formula of the Pearson correlation coefficient is:

[0070]

[0071] Among them, r represents the Pearson correlation coefficient, x i and i represents the values ​​of two variables at the i-th observation, and represents the average of the two variables in the sample, and n represents the total number of observations. The Pearson correlation coefficient r ranges from -1 to 1, where 0 represents no linear relationship, a positive value represents a positive correlation, and a negative value represents a negative correlation.

[0072] In this embodiment, the data of the comprehensive data set described in step S3 include short-term daily rainfall intensity, short-term six-day rainfall intensity, short-term seven-day rainfall intensity, long-term one-month rainfall intensity, long-term two-month rainfall intensity, daily reservoir water level fluctuations, short-term reservoir water level fluctuations, the stage of landslide deformation, the speed of landslide deformation and the acceleration of landslide deformation.

[0073] In this embodiment, the specific process of processing landslide displacement data by the CNN-LSTM machine learning model combines the advantages of convolutional neural network (CNN) and long short-term memory network (LSTM), aiming to effectively capture the local characteristics and temporal dependency of landslide displacement data; the network structure of the CNN-LSTM machine learning model includes multiple convolutional layers, pooling layers and LSTM layers connected in sequence; the multiple convolutional layers and pooling layers are used to extract the local characteristics of landslide displacement data, such as periodic changes and trend changes; the LSTM layer controls the flow of information through its gating mechanism (input gate, forget gate and output gate), thereby effectively processing the long-term dependency and gradient vanishing problems in the landslide displacement time series data; in the training process of the CNN-LSTM machine learning model, iterative training is performed using the prepared training data, and parameters such as the loss function, optimizer and evaluation index are set according to actual needs in the compilation stage of the CNN-LSTM machine learning model.

[0074] The step S5 specifically includes the following sub-steps:

[0075] S51. Input the data in the comprehensive data set into the convolution layer for key feature extraction, and output the key features; the convolution layer slides the convolution kernel on the input data, performs local connection and weight sharing operations, so as to extract the key features in the data;

[0076] S52. Input the key features into the pooling layer for downsampling, reduce the latitude of the key features, and output the features after dimensionality reduction;

[0077] S53. Input the reduced-dimensional features into the LSTM layer to capture the long-term dependency of the landslide displacement data, and output the landslide displacement prediction results.

[0078] Embodiment 2:

[0079] This embodiment provides a landslide displacement prediction system based on landslide movement mechanism and machine learning on the basis of embodiment 1, which can be used to implement the landslide displacement prediction method based on landslide movement mechanism and machine learning as in the above embodiment. The system includes:

[0080] Data acquisition module, used to collect landslide displacement data, reservoir water level data and rainfall data in the research target area;

[0081] A feature extraction module, used for receiving the landslide displacement data, and selecting a fitting function according to the landslide displacement data to extract the landslide movement feature;

[0082] A comprehensive data set construction module is used to receive the reservoir water level data, rainfall data and landslide movement characteristics to perform landslide displacement correlation analysis, and select strong correlation to construct a comprehensive data set;

[0083] Model building module, used to build CNN-LSTM machine learning model;

[0084] The landslide displacement prediction result module is used to receive the comprehensive data set, input the comprehensive data set into the CNN-LSTM machine learning model, and output the landslide displacement prediction result.

[0085] In this embodiment, an electronic device, a readable storage medium, and a computer program product are also provided.

[0086] In this embodiment, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in the above embodiment.

[0087] In this embodiment, the readable storage medium may be a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to enable a computer to execute the method according to the above embodiment.

[0088] In this embodiment, the computer program product includes a computer program, and when the computer program is executed by a processor, the method according to the above embodiment is implemented.

[0089] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0090] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0091] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0092] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0093] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0094] Embodiment 3:

[0095] On the basis of Example 1, this embodiment selects a reservoir area in City A where landslide disasters are prevalent as the study area, collects relevant data of the study area for research, and selects typical landslide displacement-time curves for research. City A is a typical mountainous city. Complex topographic and geomorphic conditions, geological conditions and frequent heavy rainfall events lead to frequent geological disasters in City A. The regular water level fluctuations in the reservoir area cause the coastal slopes to be periodically submerged in water, resulting in problems such as disordered groundwater seepage field inside the slopes, serious exchange and loss of material components, and frequent geological disasters. Therefore, the two influencing factors of rainfall and reservoir water level are mainly collected, and long-term and short-term rainfall intensity, reservoir water level fluctuations and reservoir water level periodic changes are extracted as input indicators.

[0096] In this embodiment, according to the displacement data of the reservoir research area in City A, the Verhulst inverse function model is selected for fitting, specifically:

[0097] Landslide displacement data over a period of time were collected from the landslide monitoring system of a reservoir area in City A. These data usually include time series and corresponding landslide displacement. In the data preprocessing stage, the original data was cleaned, outliers and missing values ​​were removed, and the continuity and integrity of the data were ensured. The standard form of the Verhulst inverse function model is:

[0098]

[0099] Among them, y represents the predicted value, that is, the landslide displacement, K represents the maximum possible value of the landslide displacement, y0 represents the initial value, that is, the landslide displacement at the beginning of the time series, e represents the base of the natural logarithm, r represents the rate of change of the landslide displacement over time, and t represents time;

[0100] The Verhulst inverse function model is fitted to the preprocessed landslide displacement data using a nonlinear optimization algorithm. During the fitting process, the parameters K, y0, and r are continuously adjusted to minimize the difference between the model prediction value and the actual observation value, which includes the following steps:

[0101] A1. Initialize the values ​​of parameters K, y0 and r;

[0102] A2. Use nonlinear optimization algorithm to iteratively adjust the values ​​of K, y0 and r;

[0103] A3. Calculate the error between the predicted value of the Verhulst inverse function model and the actual observed value;

[0104] A4. Determine whether the error reaches the preset convergence condition or whether the number of iterations reaches the maximum limit. If so, the process ends and the values ​​of parameters K, y0 and r are output to obtain the Verhulst inverse function model fitted to the preprocessed landslide displacement data; if not, return to step A2.

[0105] In this embodiment, the nonlinear optimization algorithm may be a nonlinear least squares method or a Levenberg-Marquardt algorithm;

[0106] The error between the predicted value of the Verhulst inverse function model and the actual observed value may be a mean square error (MSE).

[0107] In this embodiment, after the CNN-LSTM machine learning model is trained, the trained CNN-LSTM machine learning model is used to perform landslide displacement prediction verification. By inputting the verification set data, the CNN-LSTM machine learning model can output the corresponding landslide displacement prediction value to analyze the prediction accuracy and error of the model. The mean absolute percentage error (MAPE) and root mean square error (RMSE) are selected as accuracy analysis indicators to evaluate the CNN-LSTM machine learning model;

[0108] The calculation formula of the MAPE is:

[0109]

[0110] Where T represents the number of samples, y l represents the actual value of the nth sample, Represents the predicted value of the nth sample;

[0111] The calculation formula of the RMSE is:

[0112]

[0113] MAPE can reflect the relative error between the predicted value and the actual value, and is one of the important indicators for measuring the accuracy of the prediction model. By calculating the absolute error between each predicted value and the actual value, and comparing it with the actual value to obtain the percentage error, and finally averaging all the percentage errors, the MAPE value can be obtained. The smaller the index is, the higher the prediction accuracy of the model is. RMSE measures the standard deviation between the predicted value and the actual value. It takes into account the square of the error and is therefore more sensitive to large errors. The RMSE value is obtained by calculating the square root of the average of the sum of the squares of the differences between each predicted value and the actual value. Similar to MAPE, the smaller the RMSE value, the better the prediction performance of the model. When selecting these two indicators for evaluation, the overall model is firstly used for multiple landslide displacement predictions, and then the MAPE and RMSE values ​​of each prediction are calculated respectively.

[0114] In this embodiment, according to the formula: The calculated MAPE value is 5.1 on average. This means that the average absolute error of the CNN-LSTM machine learning model's predictions as a percentage of the average of the actual observations is 5.1. MAPE is a common way to measure the accuracy of a forecast because it compares the error to the actual value, providing a relatively intuitive error size. A lower MAPE value generally indicates that the model's predictions are more accurate. A MAPE value of 5.1 indicates that the CNN-LSTM machine learning model has relatively good predictive performance.

[0115] In this embodiment, according to the formula: The calculated RMSE value is 4.9 on average. This value indicates that the average standard deviation between the predicted value and the actual observed value is 4.9. RMSE is a common method to measure prediction error. It takes into account the sum of the squares of all prediction errors and then takes the square root. This method is more severe in penalizing larger errors and therefore can more sensitively reflect the prediction performance of the CNN-LSTM machine learning model. A lower RMSE value usually means that the CNN-LSTM machine learning model has a smaller prediction error and more stable prediction results. An RMSE value of 4.9 indicates that the model has a smaller prediction error and a reliable prediction result.

[0116] The two indicators, MAPE and RMSE, evaluate the CNN-LSTM machine learning model from the aspects of accuracy and stability. Among them, a lower MAPE value indicates that the model has a higher prediction accuracy, while a lower RMSE value indicates that the model's prediction results are more stable and reliable. Overall, the values ​​of these two indicators are relatively ideal, indicating that the landslide displacement prediction model has shown certain advantages in terms of accuracy, stability and reliability. Through the statistics and analysis of these indicator values, we can comprehensively evaluate the prediction performance of the model, including its accuracy, stability and reliability.

[0117] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A landslide displacement prediction method based on landslide movement mechanism and machine learning, characterized in that: The method comprises the following steps: S1. Collect landslide displacement data, reservoir water level data and rainfall data in the research target area; S2. Select a fitting function based on the landslide displacement data to extract the landslide movement characteristics; S3. Conduct landslide displacement correlation analysis on reservoir water level data, rainfall data and landslide movement characteristics, and select strong correlations to construct a comprehensive data set; S4. Build CNN-LSTM machine learning model; S5. Input the data in the comprehensive dataset into the CNN-LSTM machine learning model and output the landslide displacement prediction results.

2. The landslide displacement prediction method based on landslide movement mechanism and machine learning according to claim 1 is characterized in that: The step S2 specifically includes the following sub-steps: S21. Perform data cleaning, outlier monitoring and elimination, standardization and normalization on the landslide displacement data to obtain pre-processed landslide displacement data; S22. Draw a landslide displacement curve based on the preprocessed landslide displacement data; S23. Analyze the dynamic characteristics of the landslide displacement curve; the dynamic characteristics include overall shape, fluctuation, mutation and trend change; S24. According to the dynamic characteristic analysis results of the landslide displacement curve, a mathematical model is selected, and a fitting function is obtained by fitting using a nonlinear optimization algorithm; S25. Extracting landslide movement characteristics according to the fitting function; wherein the landslide movement characteristics include the landslide deformation stage, the landslide deformation speed and the landslide deformation acceleration; the landslide deformation stage includes the initial deformation stage, the constant speed deformation stage, the accelerated deformation stage and the possible stable or broken stage.

3. The landslide displacement prediction method based on landslide movement mechanism and machine learning according to claim 2 is characterized in that: The step S24 specifically includes the following sub-steps: S241. Select a mathematical model based on the dynamic characteristics analysis results of the landslide displacement curve; S242. Initialize the parameter values ​​of the mathematical model; S243. Iteratively updating the parameter values ​​of the mathematical model using a nonlinear optimization algorithm; S244. Calculate the error between the predicted value obtained based on the mathematical model and the actual observed value; S245. Determine whether the error reaches the preset convergence condition or whether the number of iterations reaches the maximum limit. If so, the process ends and the parameter values ​​of the updated mathematical model are output to obtain the fitting function; if not, return to step S243.

4. The landslide displacement prediction method based on landslide movement mechanism and machine learning according to claim 1 is characterized in that: In step S3, the Pearson correlation coefficient is used to perform correlation analysis; The calculation formula of the Pearson correlation coefficient is: Among them, r represents the Pearson correlation coefficient, x i and i represents the value of the two variables at the i-th observation, x and y represent the average values ​​of the two variables in the sample, and n represents the total number of observations.

5. The landslide displacement prediction method based on landslide movement mechanism and machine learning according to claim 1 is characterized in that: The data of the comprehensive data set described in step S3 include short-term daily rainfall intensity, short-term six-day rainfall intensity, short-term seven-day rainfall intensity, long-term one-month rainfall intensity, long-term two-month rainfall intensity, daily reservoir water level fluctuation, short-term reservoir water level fluctuation, stage of landslide deformation, speed of landslide deformation and acceleration of landslide deformation.

6. The method for predicting landslide displacement based on landslide movement mechanism and machine learning according to claim 1, characterized in that: The step S5 specifically includes the following sub-steps: S51. Input the data in the comprehensive data set into the convolution layer to extract key features, and output the key features; S52. Input the key features into the pooling layer for downsampling, and output the features after dimensionality reduction; S53. Input the reduced-dimensional features into the LSTM layer to capture the long-term dependency of the landslide displacement time series, and output the landslide displacement prediction results.

7. A landslide displacement prediction system based on landslide movement mechanism and machine learning, characterized in that: The system comprises: Data acquisition module, used to collect landslide displacement data, reservoir water level data and rainfall data in the research target area; A feature extraction module, used for receiving the landslide displacement data, and selecting a fitting function according to the landslide displacement data to extract the landslide movement feature; A comprehensive data set construction module is used to receive the reservoir water level data, rainfall data and landslide movement characteristics to perform landslide displacement correlation analysis, and select strong correlation to construct a comprehensive data set; Model building module, used to build CNN-LSTM machine learning model; The landslide displacement prediction module is used to receive the comprehensive data set, input the comprehensive data set into the CNN-LSTM machine learning model, and output the landslide displacement prediction result.

8. An electronic device, characterized in that: include: at least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor executes the method described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute the method according to any one of claims 1-6.

10. A computer program product, characterized in that The method comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images