Landslide displacement prediction method and system based on sliding window method and stacking ensemble learning algorithm

Abstract

The application discloses a landslide displacement prediction method and system based on a sliding window method and a stacking integrated learning algorithm, and comprises the following steps: collecting landslide displacement data of each monitoring point, and adopting a moving average method to decompose the data into trend item data and periodic item data; constructing a data set according to the data, and preliminarily screening candidate input factors; performing Pearson correlation coefficient analysis on the candidate input factors and the periodic item data to obtain model input factors; decomposing the model input factors into a fitting set and a prediction set, and performing prediction based on the sliding window method and the stacking integrated deep learning algorithm to obtain periodic item displacement prediction results; performing prediction on the trend item data by using a support vector regression algorithm to obtain trend item displacement prediction results; and adding the periodic item displacement prediction results and the trend item displacement prediction results to obtain total landslide displacement prediction results; and the stacking integrated model adopted by the application effectively fuses multiple deep learning algorithms to predict landslide displacement, and further improves the accuracy and stability of the prediction results.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of landslide displacement prediction, in particular to a landslide displacement prediction method and system based on a sliding window method and a Stacking integrated learning algorithm. BACKGROUND

[0002] The impact of landslides on human production and life cannot be ignored, which not only threatens human life, but also has destructive effects on property and the environment. Therefore, it is of great theoretical significance and practical value to carry out landslide displacement prediction research to reduce the harm caused by landslides.

[0003] At present, high-precision prediction of landslide displacement has important reference value for landslide prediction and early warning, which has attracted the attention of scholars at home and abroad, and the development of landslide displacement prediction has been steadily carried out. However, the traditional landslide displacement prediction model often has the problem of low prediction accuracy; a single algorithm model may have the problems of being easily trapped in local optimal value and low convergence accuracy. SUMMARY

[0004] This section aims to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the above problems, the present application is proposed.

[0006] In a first aspect of the embodiments of the present application, a landslide displacement prediction method based on a sliding window method and a Stacking integrated learning algorithm is provided, comprising: collecting landslide displacement data of each monitoring point, and decomposing the landslide displacement data into trend item data and periodic item data using a moving average method; constructing a data set according to the landslide displacement data, and preliminarily screening candidate input factors, and performing Pearson correlation coefficient analysis on the candidate input factors and the periodic item data to obtain model input factors; decomposing the model input factors into a fitting set and a prediction set, and performing prediction based on a sliding window method and a Stacking integrated deep learning algorithm to obtain a periodic item displacement prediction result; predicting the trend item data using a support vector regression algorithm to obtain a trend item displacement prediction result, and adding the periodic item displacement prediction result and the trend item displacement prediction result to obtain a total landslide displacement prediction result.

[0007] As a preferred scheme of the landslide displacement prediction method based on the sliding window method and the Stacking integrated learning algorithm, the acquisition of the periodic item displacement prediction result comprises,

[0008] Input the model input factors into the original data set, and decompose the original data set into a fitting set S and a prediction set P according to a certain ratio;

[0009] The fitting set S is divided into n+1 subsets S1, S2, ..., S n+1 , select the first i subsets as the training set to train the base model, use the trained base model to verify the i+1th data subset, and output the verification results. Repeat this process n times to obtain all the verification results of the base model;

[0010] All verification results are output as new features. m base models are verified m times to obtain m new features, that is, m new fitting sets. The new fitting sets are then combined to obtain a fitting set N, which is used as the new fitting set input of the second layer.

[0011] The prediction set P is trained using the trained base model to obtain a prediction result, which is then weighted and used as the new prediction set input for the second layer.

[0012] As a preferred solution of the landslide displacement prediction method based on the sliding window method and the Stacking ensemble learning algorithm described in the present invention, it also includes:

[0013] In the second-level algorithm, the fitting set N is divided into s subsets, the first s-1 subsets are used for meta-model training, and the remaining subsets are used for meta-model parameter optimization. The new prediction set is then predicted to obtain the final periodic item displacement prediction result.

[0014] As a preferred solution of the landslide displacement prediction method based on the sliding window method and the Stacking ensemble learning algorithm described in the present invention, wherein: weighted processing of the prediction results includes:

[0015] The basis for weight distribution is the mean absolute error of each base model, which is calculated as follows:

[0016]

[0017] Among them, MAE represents the mean absolute error, m represents the number of base models, and y i Indicates the actual value, represents the predicted value;

[0018] Set the prediction result output by each base model to C m , the weight coefficient of each result is G m , then the final total output result is expressed as ∑C m G m .

[0019] As a preferred scheme of the landslide displacement prediction method based on the sliding window method and the Stacking integrated learning algorithm, the acquisition of the trend item displacement prediction result comprises,

[0020] According to the trend item data, 60%-70% of the data are trained as a training set of the trend item prediction model, and 10%-20% of the data are used for parameter optimization of the trend item prediction model.

[0021] The trained trend item prediction model is used to predict the remaining data to obtain the trend item displacement prediction result.

[0022] As a preferred scheme of the landslide displacement prediction method based on the sliding window method and the Stacking integrated learning algorithm, the adoption of the moving average method to decompose the landslide displacement data into trend item data and periodic item data comprises,

[0023] The landslide displacement data comprises original landslide displacement, daily rainfall, and daily reservoir water level.

[0024] The moving average method is adopted to smooth the fluctuation in the landslide displacement data, and the trend item data is extracted, and the periodic item data is extracted according to the moving average period.

[0025] The calculation of the moving average method comprises,

[0026]

[0027] wherein, represents the value of the trend item displacement at time step t, X t represents the value of the cumulative displacement at time step t, and n represents the moving average period.

[0028] As a preferred scheme of the landslide displacement prediction method based on the sliding window method and the Stacking integrated learning algorithm, the acquisition of the model input factor comprises,

[0029] Pearson correlation coefficient analysis is performed on the candidate input factors and the periodic item data, and a plurality of candidate input factors corresponding to the maximum Pearson correlation coefficient are selected as input factors in the periodic item displacement prediction model.

[0030] The calculation of the Pearson correlation coefficient comprises,

[0031]

[0032] wherein, X and Y represent two equal-length vectors, and N represents the number of vector elements.

[0033] A second aspect of an embodiment of the present invention provides a landslide displacement prediction system based on a sliding window method and a stacking ensemble learning algorithm, comprising:

[0034] A data acquisition unit is used to collect landslide displacement data of each monitoring point and decompose the landslide displacement data into trend item data and period item data using a moving average method;

[0035] A data processing unit is used to construct a data set based on the landslide displacement data, preliminarily screen candidate input factors, and perform Pearson correlation coefficient analysis on the candidate input factors and the periodic item data to obtain model input factors;

[0036] The displacement prediction unit is used to decompose the model input factors into a fitting set and a prediction set, and to predict based on the sliding window method and the Stacking integrated deep learning algorithm to obtain a periodic item displacement prediction result, to predict the trend item data using a support vector regression algorithm to obtain a trend item displacement prediction result, and to add the periodic item displacement prediction result and the trend item displacement prediction result to obtain a total landslide displacement prediction result.

[0037] According to a third aspect of an embodiment of the present invention, a device is provided, comprising:

[0038] processor;

[0039] a memory for storing processor-executable instructions;

[0040] The processor is configured to call the instructions stored in the memory to execute the method described in any embodiment of the present invention.

[0041] According to a fourth aspect of the embodiments of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, including:

[0042] When the computer program instructions are executed by a processor, the method according to any embodiment of the present invention is implemented.

[0043] Beneficial effects of the present invention: The present invention provides a landslide displacement prediction method and system based on the sliding window method and the Stacking ensemble learning algorithm. When predicting landslide displacement, the sliding window method is used for verification instead of simple cross-validation, which has met the requirements of logical rationality. In addition, the Stacking ensemble model adopted by the present invention can effectively integrate multiple deep learning algorithms to predict landslide displacement, further improving the accuracy and stability of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments description will be briefly introduced as follows. Obviously, the drawings in the following description only represent some of the embodiments of the present application, and all other drawings obtained by those skilled in the art without creative labor on the basis of these drawings should also belong to the protection scope of the present application.

[0045] Figure 1 The overall flowchart of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure.

[0046] Figure 2 The periodic term displacement prediction flowchart of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure.

[0047] Figure 3 The schematic diagram of the output and input factors of the fitting centralized training set and the validation set of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure.

[0048] Figure 4 The landslide displacement trend term prediction result curve of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure.

[0049] Figure 5 The slope displacement periodic term prediction result curve of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure.

[0050] Figure 6 The landslide total displacement prediction result curve of the landslide displacement prediction method and system based on the sliding window method and the stacking integrated learning algorithm provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail in combination with the drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0052] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0054] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0055] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0056] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0057] Example 1

[0058] Reference Figures 1 to 3 In one embodiment of the present invention, a landslide displacement prediction method based on a sliding window method and a stacking ensemble learning algorithm is provided, comprising:

[0059] S1: Collect the landslide displacement data of each monitoring point and use the moving average method to decompose the landslide displacement data into trend item data and period item data.

[0060] Landslide displacement data include the original displacement of the landslide, the rainfall on that day, and the reservoir water level on that day;

[0061] The moving average method is used to smooth the fluctuations in the landslide displacement data and extract the trend item data. The period item data is extracted according to the moving average period.

[0062] It should be noted that the calculation of the moving average method includes:

[0063]

[0064] in, represents the displacement value of the trend term at time step t, X t represents the value of the cumulative displacement at time step t, and n represents the moving average period;

[0065] It should be noted that the single-point moving average method is to calculate the moving average of time series data at a certain length, which can better reflect the trend and changes of the time series.

[0066] S2: Construct a data set based on the landslide displacement data, preliminarily screen candidate input factors, and perform Pearson correlation coefficient analysis on the candidate input factors and periodic item data to obtain the model input factors. It should be noted that:

[0067] Perform Pearson correlation coefficient analysis on candidate input factors and periodic item data, and select several candidate input factors corresponding to the maximum Pearson correlation coefficient as input factors in the periodic item displacement prediction model;

[0068] Specifically, the calculation of the Pearson correlation coefficient includes:

[0069]

[0070] Where X and Y represent two vectors of equal length, and N represents the number of vector elements;

[0071] It should be noted that the Pearson correlation coefficient method stipulates that the value range of the relationship between two continuous variables is [-1, 1], which is used to measure the size and direction of the relationship between two continuous variables. 1 indicates positive correlation, 0 indicates no correlation, and -1 indicates negative correlation.

[0072] S3: Decompose the model input factors into a fitting set and a prediction set, and perform prediction based on the sliding window method and stacking integrated deep learning algorithm to obtain the periodic term displacement prediction results. It should be noted that:

[0073] like Figure 2 As shown, the acquisition of the periodic term displacement prediction results includes:

[0074] Input the model input factors into the original data set, decompose the original data set into the fitting set S and the prediction set P according to a certain ratio, and divide the fitting set S into n+1 subsets S1, S2, ..., S n+1, select the first i subsets as the training set to train the base model, use the trained base model to verify the i+1th data subset, and output the verification results. Repeat this process n times to obtain all the verification results of the base model;

[0075] All verification results are output as new features. m base models are verified m times to obtain m new features, that is, m new fitting sets. Then the new fitting sets are combined to obtain a fitting set N, which is used as the new fitting set input of the second layer.

[0076] The trained base model is used to train the prediction set P to obtain the prediction results, which are then weighted and used as the new prediction set input for the second layer.

[0077] like Figure 3 As shown in the figure, in the second-level algorithm, the fitting set N is divided into s subsets, the first s-1 subsets are used for meta-model training, and the remaining subsets are used for meta-model parameter optimization. The new prediction set is then predicted to obtain the final periodic term displacement prediction result.

[0078] Furthermore, weighted processing of the prediction results includes:

[0079] The basis for weight distribution is the mean absolute error of each base model, which is calculated as follows:

[0080]

[0081] Among them, MAE represents the mean absolute error, m represents the number of base models, and y i Indicates the actual value, represents the predicted value;

[0082] Set the prediction result output by each base model to C m , the weight coefficient of each result is G m , then the final total output result is expressed as ∑C m G m .

[0083] S4: Use the support vector regression algorithm to predict the trend item data to obtain the trend item displacement prediction result, add the period item displacement prediction result and the trend item displacement prediction result to obtain the total landslide displacement prediction result. It should be noted that:

[0084] The acquisition of trend item displacement prediction results includes:

[0085] According to the trend item data, 60% to 70% of the data is used as the training set of the trend item prediction model for training, and 10% to 20% of the data is used for parameter optimization of the trend item prediction model;

[0086] The trained trend item prediction model is used to predict the remaining data to obtain the trend item displacement prediction results.

[0087] It should be noted that the present invention provides a landslide displacement prediction method and system based on the sliding window method and the Stacking ensemble learning algorithm. When predicting landslide displacement, the sliding window method is used for verification instead of simple cross-validation, which has met the requirements of logical rationality. In addition, the Stacking ensemble model adopted by the present invention can effectively integrate multiple deep learning algorithms to predict landslide displacement, further improving the accuracy and stability of the prediction results.

[0088] The second aspect of the present invention is disclosed.

[0089] Provides a landslide displacement prediction system based on the sliding window method and Stacking ensemble learning algorithm, including:

[0090] The data acquisition unit is used to collect the landslide displacement data of each monitoring point and decompose the landslide displacement data into trend item data and period item data using the moving average method;

[0091] The data processing unit is used to construct a data set based on the landslide displacement data, preliminarily screen candidate input factors, and perform Pearson correlation coefficient analysis on the candidate input factors and periodic item data to obtain model input factors;

[0092] The displacement prediction unit is used to decompose the model input factors into a fitting set and a prediction set, and to obtain the periodic displacement prediction results based on the sliding window method and the Stacking integrated deep learning algorithm. The trend item data is predicted using the support vector regression algorithm to obtain the trend item displacement prediction results. The periodic item displacement prediction results and the trend item displacement prediction results are added together to obtain the total landslide displacement prediction results.

[0093] The third aspect of the present invention is disclosed.

[0094] Provided is a device comprising:

[0095] processor;

[0096] a memory for storing processor-executable instructions;

[0097] The processor is configured to call instructions stored in the memory to execute any one of the aforementioned methods.

[0098] The fourth aspect of the present invention is disclosed.

[0099] A computer-readable storage medium is provided, on which computer program instructions are stored, including:

[0100] When the computer program instructions are executed by a processor, any of the above methods is implemented.

[0101] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0102] A computer-readable storage medium can be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: 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), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination thereof. As used herein, a computer-readable storage medium is not to be construed as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through an electrical wire.

[0103] Example 2

[0104] Reference Figures 4-6 This is the second embodiment of the present invention. Different from the first embodiment, this embodiment provides a landslide displacement prediction method and system verification test based on the sliding window method and Stacking ensemble learning algorithm to verify and illustrate the technical effects adopted in this method.

[0105] This example selects a landslide in Zigui County of the Three Gorges Reservoir area of ​​the Yangtze River and uses the method provided by the present invention to predict the landslide displacement.

[0106] First, rainfall and reservoir water level are the primary factors influencing landslide deformation. Regarding data collection, data related to landslide displacement was collected at each monitoring point. The data included the original landslide displacement, daily rainfall, and daily reservoir water level. In this example, a total of 84 monitoring time steps from January 2004 to December 2010 were selected as the research object. The trend term input factor was the trend term displacement difference. Alternative input factors for the periodic term included monthly rainfall, cumulative rainfall over the past two months, maximum daily rainfall over the month, number of days with rainfall over the month, maximum number of days with continuous rainfall over the month, average reservoir water level over the month, reservoir water level change over the month, reservoir water level change over the past two months, number of days with reservoir water level drop over the month, cumulative reservoir water level drop over the month, number of days with reservoir water level rise over the month, and cumulative reservoir water level rise over the month.

[0107] Secondly, the moving average method is used to decompose the landslide displacement data into trend items and displacement items, so as to better determine the period of the time series and its influencing factors. According to the landslide displacement trend item data in this embodiment, the first 60 groups of data are first used as training sets in the trend item prediction model for training, and then 12 groups are used for parameter optimization of the trend item prediction model. Finally, the trained trend item prediction model is used to predict the remaining 12 groups of data. The obtained trend item displacement is used as the displacement result X1 of the trend item. The landslide displacement trend item prediction result curve is shown in FIG. Figure 4 shown.

[0108] Then, the original data set is constructed based on the landslide displacement data. Based on the original data set, the candidate input factors that are beneficial to training and prediction are preliminarily screened out. The Pearson correlation coefficient analysis is performed on the periodic displacement and the candidate input factors, and the candidate input factors with high correlation are selected as the input factors. In the embodiment, 84 groups of data are input into the original data set, the first 72 groups of data are used as fitting sets, and the remaining 12 groups of data are used as prediction sets. The fitting sets are divided into 6 subsets S1, S2, S3, S4, S5, and S6. The first i subsets are selected in turn as training sets to train the base model to obtain a trained base model. The trained base model is used to verify the i+1 data subset, and the verification result i (i=1, 2, 3, 4, 5) is output. This is repeated 5 times to obtain 5 groups of verification results of the base model, which are output as new features. After all 5 base models are verified, 5 new features are obtained, which are used as new inputs for the second layer and combined into a new fitting set; the new fitting set is divided into 5 subsets, the first 4 subsets are used for training the meta-model, and the remaining subsets are used for parameter optimization of the meta-model. The new prediction set is then predicted to obtain the final landslide period item prediction result X2. The landslide displacement period item prediction result curve is shown in FIG. Figure 5 shown.

[0109] It should be noted that in the weighted calculation, the prediction results output by the five base models under each algorithm in the first layer are recorded as G1, G2, G3, G4, and G5, the MAE indexes of the five base models are recorded as M1, M2, M3, M4, and M5, and the weight coefficients of the five base models are recorded as C1, C2, C3, C4, and C5, respectively. The following calculation results are obtained:

[0110]

[0111] The final prediction result of the periodic term is X2=C1G1+C2G2+C3G3+C4G4+C5G5. The displacement prediction results of the periodic term and the displacement prediction results of the trend term are added together to obtain the total landslide displacement prediction result as shown in the following figure: Figure 6 The predicted result data are shown in Table 1.

[0112] Table 1: Prediction results data.

[0113] Model Trend item RMSE (mm) Periodic term RMSE (mm) Total displacement RMSE (mm) Stacking 2.30 27.49 29.23 LSTM 2.30 29.27 30.77 CNN 2.30 34.01 35.77 RNN 2.30 30.41 31.36 DNN 2.30 32.06 33.05

[0114] Depend on Figure 6 As shown in Table 1, overall, the landslide displacement prediction model based on the sliding window method and Stacking integrated deep learning algorithm is better than other deep learning landslide displacement prediction models.

[0115] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. The landslide displacement prediction method based on the sliding window method and Stacking ensemble learning algorithm is characterized by: include: Collecting landslide displacement data from each monitoring point, and decomposing the landslide displacement data into trend item data and period item data using a moving average method; Constructing a data set based on the landslide displacement data, preliminarily screening candidate input factors, and performing Pearson correlation coefficient analysis on the candidate input factors and the periodic item data to obtain model input factors; The model input factors are decomposed into a fitting set and a prediction set, and predictions are performed based on a sliding window method and a stacking integrated deep learning algorithm to obtain the periodic term displacement prediction results; Using a support vector regression algorithm to predict the trend item data to obtain a trend item displacement prediction result, and adding the period item displacement prediction result and the trend item displacement prediction result to obtain a total landslide displacement prediction result; The acquisition of the periodic term displacement prediction result includes: The model input factors are input into the original data set, and the original data set is decomposed into fitting sets according to a certain ratio. S With the prediction set P ; The fitting set S Divided equally n +1 subset , select the front i The subset is used as the training set to train the base model, and the trained base model is used to train the i +1 data subset for verification and output the verification results, and repeat this process n times, obtaining all verification results of the base model; All verification results are output as new features. m base models are verified m times to obtain m new features, that is, m new fitting sets. The new fitting sets are then combined to obtain a fitting set N, which is used as the new fitting set input of the second layer. Use the trained base model to predict the P The training is performed to obtain prediction results, which are then weighted and used as the new prediction set input for the second layer.

2. The landslide displacement prediction method based on the sliding window method and the stacking ensemble learning algorithm according to claim 1 is characterized in that: Also includes, In the second-level algorithm, the fitting set N is divided into s subsets, the first s-1 subsets are used for meta-model training, and the remaining subsets are used for meta-model parameter optimization. The new prediction set is then predicted to obtain the final periodic item displacement prediction result.

3. The landslide displacement prediction method based on the sliding window method and the stacking ensemble learning algorithm according to claim 2 is characterized in that: The weighted processing of the prediction results includes: The basis for weight distribution is the mean absolute error of each base model, which is calculated as follows: in, represents the mean absolute error, m represents the number of base models, Indicates the actual value, represents the predicted value; Set the prediction result output by each base model to , the weight coefficient of each result is , then the final total output result is expressed as .

4. The landslide displacement prediction method based on the sliding window method and the stacking ensemble learning algorithm according to claim 3 is characterized in that: The acquisition of the trend item displacement prediction result includes: Based on the trend item data, 60% to 70% of the data is used as a training set for the trend item prediction model, and 10% to 20% of the data is used for parameter optimization of the trend item prediction model; The trained trend item prediction model is used to predict the remaining data to obtain the trend item displacement prediction results.

5. The landslide displacement prediction method based on the sliding window method and the stacking ensemble learning algorithm according to claim 1 or 4, characterized in that: The use of the moving average method to decompose the landslide displacement data into trend item data and period item data includes: The landslide displacement data includes the original displacement of the landslide, the rainfall on that day, and the reservoir water level on that day; Using a moving average method to smooth the fluctuations in the landslide displacement data, and extracting trend item data, and extracting period item data according to the moving average period; The calculation of the moving average method includes: in, Indicates the displacement of the trend term in time step t The value of Indicates the cumulative displacement in time step t The value of n Represents the moving average period.

6. The landslide displacement prediction method based on the sliding window method and the stacking ensemble learning algorithm according to claim 5 is characterized in that: The acquisition of the model input factors includes: Performing a Pearson correlation coefficient analysis on the candidate input factors and the periodic item data, and selecting several candidate input factors corresponding to the maximum values ​​of the Pearson correlation coefficients as input factors in the periodic item displacement prediction model; The calculation of the Pearson correlation coefficient includes, in, X and Y represents two vectors of equal length, N Indicates the number of vector elements.

7. A landslide displacement prediction system based on a sliding window method and a stacking ensemble learning algorithm, wherein the landslide displacement prediction method based on a sliding window method and a stacking ensemble learning algorithm according to any one of claims 1 to 6 is applied, characterized in that: include: A data acquisition unit is used to collect landslide displacement data of each monitoring point and decompose the landslide displacement data into trend item data and period item data using a moving average method; A data processing unit is used to construct a data set based on the landslide displacement data, preliminarily screen candidate input factors, and perform Pearson correlation coefficient analysis on the candidate input factors and the periodic item data to obtain model input factors; The displacement prediction unit is used to decompose the model input factors into a fitting set and a prediction set, and to predict based on the sliding window method and the Stacking integrated deep learning algorithm to obtain a periodic item displacement prediction result, to predict the trend item data using a support vector regression algorithm to obtain a trend item displacement prediction result, and to add the periodic item displacement prediction result and the trend item displacement prediction result to obtain a total landslide displacement prediction result.

8. A device, characterized in that The device comprises, processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

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

Images