Rainfall-induced landslide early warning methods and devices based on machine learning models and numerical analysis

Abstract

This application proposes a rainfall-based landslide early warning method using machine learning models and numerical analysis, comprising: constructing a landslide displacement prediction model based on displacement data and rainfall data from apparent landslide monitoring points, and obtaining landslide displacement prediction curves under different rainfall scenarios based on the model; determining the landslide extent based on geological surveys and obtaining soil mechanical parameters through geotechnical tests; selecting a calculation profile based on the landslide extent, and performing two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain a plastic zone distribution map and a displacement distribution cloud map; calculating the displacement threshold at the corresponding apparent monitoring points when the plastic zone of the slip zone evolves to different penetration rates based on the plastic zone distribution map and displacement cloud map; determining the time for displacement evolution to the threshold displacement based on the displacement threshold combined with the landslide displacement prediction curves, and implementing landslide safety early warning based on the evolution time and early warning level. This invention, employing the above scheme, has significant theoretical and engineering value for landslide displacement prediction and safety early warning.

Description

Technical Field

[0001] This application relates to the fields of landslide prevention and mitigation and geotechnical engineering technology, and in particular to a method and device for early warning of landslides caused by rainfall using machine learning models and numerical analysis. Background Technology

[0002] With the continuous development of hydropower resources in Southwest my country, a large number of large river-type reservoirs have been formed. Numerous water-related slopes within these reservoir areas have gradually deformed and deteriorated, leading to landslides and surges. Individual landslides and their secondary disasters, such as environmental pollution, surges generated by landslides, and barrier lakes, seriously threaten the lives and property of people in the reservoir areas and the stable operation of the reservoirs. Therefore, landslides have become one of the major geological hazards in Southwest my country.

[0003] Currently, landslide displacement prediction and safety early warning methods are divided into numerical simulation calculations based on physical and mechanical foundations and data-driven prediction models based on safety monitoring data. Numerical analysis-based landslide displacement prediction and early warning methods often involve simplification of geological models and mechanical parameters, and the modeling and calculation processes are relatively time-consuming and labor-intensive. Data-driven models based on statistical and machine learning algorithms are significantly affected by the model structure. While global optimization of model parameters can achieve relatively accurate displacement predictions, the physical meaning of the model is unclear, and it is impossible to determine early warning thresholds using such models. Therefore, relying solely on physical and mechanical models or data-driven models cannot adequately reflect the physical and mechanical characteristics of landslide displacement evolution in relation to external influencing factors, and it is difficult to determine early warning indicators and thresholds with clear physical meanings. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the first objective of this application is to propose a rainfall-based landslide early warning method using machine learning models and numerical analysis. This method solves the technical problem that existing methods are inefficient and cannot accurately predict and warn of landslide displacement. It enables accurate prediction of landslide displacement and early warning of landslide safety, providing strong support for the construction of safety monitoring and early warning systems for reservoir bank landslides and slopes under complex geological conditions. This method has significant application value and engineering significance.

[0006] The second objective of this application is to propose a rainfall-based landslide early warning device based on machine learning models and numerical analysis.

[0007] To achieve the above objectives, the first aspect of this application proposes a rainfall-based landslide early warning method using machine learning models and numerical analysis, comprising: constructing a landslide displacement prediction model based on displacement data from apparent landslide monitoring points and rainfall data, and obtaining landslide displacement prediction curves under different rainfall scenarios based on the prediction model; determining the landslide extent based on geological surveys, and obtaining soil mechanical parameters through geotechnical tests and parameter inversion; selecting a calculation profile based on the landslide extent, and performing two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain a plastic zone distribution map and a displacement distribution cloud map; calculating the displacement threshold at the apparent monitoring points when the plastic zone of the slip zone evolves to different penetration rates based on the plastic zone distribution map and the displacement cloud map; determining the time for the displacement to evolve to the threshold displacement based on the displacement threshold at the apparent monitoring points and the landslide displacement prediction curves under different rainfall scenarios, and implementing landslide safety early warning based on the evolution time and early warning level.

[0008] The rainfall-induced landslide early warning method based on machine learning models and numerical analysis in this application, through machine learning prediction models and numerical analysis, and based on the plastic zone penetration rate analyzed by numerical simulation, scientifically proposes a landslide displacement threshold according to the physical process and mechanical mechanism of landslide failure, and then reasonably determines the landslide early warning time interval. This has important theoretical value and engineering significance for landslide displacement prediction and safety early warning.

[0009] Optionally, in one embodiment of this application, the displacement data is a displacement time series, and the rainfall data is an external rainfall series. The external rainfall series is a sequence composed of rainfall characteristics, including cumulative rainfall, rainfall intensity, and rainfall duration. A landslide displacement prediction model is constructed based on the displacement data and rainfall data from the apparent monitoring points of the landslide, including:

[0010] A training set was constructed based on displacement time series and external rainfall series.

[0011] The external rainfall sequence and displacement time series are used as inputs to the support vector machine model, and the displacement increment is used as the output of the support vector machine model. The support vector machine model is trained using a training set to obtain a landslide displacement prediction model. The kernel parameters of the support vector machine model are globally optimized by the particle swarm optimization algorithm.

[0012] Optionally, in one embodiment of this application, the external rainfall sequence is represented as:

[0013] P = {P} a P max P t}

[0014] Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall;

[0015] The displacement increment is expressed as:

[0016] Δx = SVM(D, P) a ,P max ,P t )

[0017] Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

[0018] Optionally, in one embodiment of this application, the rainfall scenario is determined by rainfall characteristics, and landslide displacement prediction curves under different rainfall scenarios are obtained based on a prediction model, including:

[0019] By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

[0020] Optionally, in one embodiment of this application, the soil mechanical parameters include internal cohesion and internal friction angle, the landslide range includes landslide surface displacement, and a two-dimensional numerical analysis of the calculated profile is performed based on the soil mechanical parameters, including:

[0021] Two-dimensional numerical analysis of the calculated profile was performed using continuum mechanics analysis software to obtain the profile displacement and plastic zone.

[0022] The internal cohesion and internal friction angle were gradually reduced proportionally using the strength reduction method. Numerical calculations were performed based on the selected penetration rates of different plastic zones of the slip zone to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

[0023] To achieve the above objectives, a second aspect of the present invention provides a rainfall-based landslide early warning device based on machine learning models and numerical analysis, comprising a construction module, a first calculation module, a second calculation module, a third calculation module, and an early warning module, wherein...

[0024] The module is used to build a landslide displacement prediction model based on displacement data and rainfall data from apparent landslide monitoring points, and to obtain landslide displacement prediction curves under different rainfall scenarios based on the prediction model.

[0025] The first calculation module is used to determine the landslide range based on geological surveys and to obtain soil mechanical parameters through geotechnical tests and calculation parameter inversion.

[0026] The second calculation module is used to select the calculation profile according to the landslide range, and perform two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain the plastic zone distribution map and displacement distribution cloud map.

[0027] The third calculation module is used to calculate the displacement threshold at the corresponding apparent monitoring point when the plastic zone of the sliding strip evolves to different penetration rates, based on the plastic zone distribution map and displacement cloud map.

[0028] The early warning module is used to determine the time for the displacement to evolve to the threshold displacement based on the displacement threshold of the apparent monitoring points and the landslide displacement prediction curve under different rainfall scenarios, and to implement landslide safety early warning based on the evolution time and early warning level.

[0029] Optionally, in one embodiment of this application, the displacement data is a displacement time series, and the rainfall data is an external rainfall series. The external rainfall series is a sequence composed of rainfall characteristics, including cumulative rainfall, rainfall intensity, and rainfall duration. A landslide displacement prediction model is constructed based on the displacement data and rainfall data from the apparent monitoring points of the landslide, including:

[0030] A training set was constructed based on displacement time series and external rainfall series.

[0031] The external rainfall sequence and displacement time series are used as inputs to the support vector machine model, and the displacement increment is used as the output of the support vector machine model. The support vector machine model is trained using a training set to obtain a landslide displacement prediction model. The kernel parameters of the support vector machine model are globally optimized by the particle swarm optimization algorithm.

[0032] Optionally, in one embodiment of this application, the external rainfall sequence is represented as:

[0033] P = {P} a P max P t}

[0034] Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall;

[0035] The displacement increment is expressed as:

[0036] Δx = SVM(D, P) a ,P max ,P t )

[0037] Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

[0038] Optionally, in one embodiment of this application, the rainfall scenario is determined by rainfall characteristics, and landslide displacement prediction curves under different rainfall scenarios are obtained based on a prediction model, including:

[0039] By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

[0040] Optionally, in one embodiment of this application, the soil mechanical parameters include internal cohesion and internal friction angle, the landslide range includes landslide surface displacement, and a two-dimensional numerical analysis of the calculated profile is performed based on the soil mechanical parameters, including:

[0041] Two-dimensional numerical analysis of the calculated profile was performed using continuum mechanics analysis software to obtain the profile displacement and plastic zone.

[0042] The internal cohesion and internal friction angle were gradually reduced proportionally using the strength reduction method. Numerical calculations were performed based on the selected penetration rates of different plastic zones of the slip zone to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

[0043] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0044] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0045] Figure 1 This is a flowchart illustrating a rainfall-based landslide early warning method based on a machine learning model and numerical analysis, as provided in Embodiment 1 of this application.

[0046] Figure 2 This is a graph showing the landslide displacement prediction sequence under different rainfall scenarios based on PSO-SVM, according to an embodiment of this application.

[0047] Figure 3 This is a distribution diagram of the plastic zone with different plastic zone penetration rates corresponding to the typical calculation profile of the embodiment of this application;

[0048] Figure 4 Displacement cloud diagrams corresponding to different plastic zone penetration rates for typical calculation cross-sections in embodiments of this application;

[0049] Figure 5 A schematic diagram illustrating the determination of landslide evolution time intervals by combining landslide predicted displacement sequences and corresponding displacement thresholds in an embodiment of this application;

[0050] Figure 6This is a schematic diagram of the structure of a rainfall-induced landslide early warning device based on a machine learning model and numerical analysis, provided in an embodiment of this application. Detailed Implementation

[0051] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0052] The following description, with reference to the accompanying drawings, describes a rainfall-based landslide early warning method and apparatus based on machine learning models and numerical analysis, representing embodiments of this application.

[0053] Figure 1 This is a flowchart illustrating a rainfall-based landslide early warning method based on a machine learning model and numerical analysis, as provided in Embodiment 1 of this application.

[0054] like Figure 1 As shown, the rainfall-based landslide early warning method based on the machine learning model and numerical analysis includes the following steps:

[0055] Step 101: Construct a landslide displacement prediction model based on displacement data and rainfall data from apparent landslide monitoring points, and obtain landslide displacement prediction curves under different rainfall scenarios based on the prediction model.

[0056] Step 102: Determine the landslide area based on geological surveys, and obtain soil mechanical parameters through geotechnical tests and parameter inversion.

[0057] Step 103: Select the calculation profile according to the landslide range, and perform two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain the plastic zone distribution map and displacement distribution cloud map.

[0058] Step 104: Based on the plastic zone distribution map and displacement cloud map, calculate the displacement threshold at the corresponding apparent monitoring point when the plastic zone of the sliding strip evolves to different penetration rates.

[0059] Step 105: Based on the displacement threshold of the apparent monitoring points and the landslide displacement prediction curves under different rainfall scenarios, determine the time for the displacement to evolve to the threshold displacement, and implement landslide safety early warning based on the evolution time and early warning level.

[0060] The rainfall-induced landslide early warning method based on machine learning models and numerical analysis in this application, through machine learning prediction models and numerical analysis, and based on the plastic zone penetration rate analyzed by numerical simulation, scientifically proposes a landslide displacement threshold according to the physical process and mechanical mechanism of landslide failure, and then reasonably determines the landslide early warning time interval. This has important theoretical value and engineering significance for landslide displacement prediction and safety early warning.

[0061] Optionally, in one embodiment of this application, the displacement data is a displacement time series D based on GNSS apparent monitoring points arranged on the surface of the landslide (here, the landslide is an accumulated landslide or an engineering slope, etc. that has already slid), and the rainfall data is an external rainfall series P obtained from rain gauges around the landslide during the corresponding monitoring period. The external rainfall series P is a sequence composed of rainfall characteristics, including cumulative P. a Rainfall intensity P max and duration of rainfall P t To analyze the correlation between landslide displacement sequence D and rainfall characteristics P, we first obtained time-series displacement data by monitoring the apparent deformation of the landslide body. Combined with rainfall information, we can roughly conclude that the deformation is large when the rainfall is heavy and relatively small when the rainfall is light.

[0062] Based on displacement data and rainfall data from apparent landslide monitoring points, a suitable machine learning model is constructed to predict specific landslide displacements, including:

[0063] Select the cumulative rainfall P in the previous period a Previous rainfall intensity P max Duration of previous rainfall (P) t The displacement time series D is used as the input to the prediction model (PSO-SVM model), and the displacement increment Δx is the output of the prediction model. The data is divided into a model training set and a model validation training set at a ratio of 80% and 20% respectively. The parameters of the support vector machine kernel function are determined by the particle swarm optimization (PSO) method. The optimization criterion is that the sum of squared errors between the predicted and actual values ​​approaches zero. The PSO-SVM model is established to determine the nonlinear mapping relationship between the landslide displacement increment and external influencing factors: Δx = SVM(D, P...). a P max P t ).

[0064] Taking two rainfall scenarios (rainfall scenario 1 and rainfall scenario 2) as examples, the cumulative rainfall, rainfall intensity and rainfall duration under the corresponding rainfall scenarios are selected as model inputs. Based on the PSO-SVM model constructed above, landslide displacement prediction is realized. After determining the input and output of the support vector machine (SVM), samples are used to train its kernel function parameters. Then, the optimized parameters are substituted into the support vector machine model for prediction.

[0065] Figure 2 The graphs show the displacement versus time for rainfall scenarios 1 and 2. Figure 2 The horizontal axis represents time (in weeks), and the vertical axis represents displacement (in mm).

[0066] Different rainfall scenarios are selected here, but multiple scenarios can also be chosen. For the displacement at different times, the input to the Support Vector Machine (SVM) model is the rainfall scenario (previous cumulative rainfall, previous maximum daily rainfall, and number of consecutive rainy days) and the previous displacement time series. The output is the displacement increment; the current displacement plus the displacement increment equals the current displacement.

[0067] Optionally, in one embodiment of this application, the external rainfall sequence is represented as:

[0068] P = {P} a P max P t}

[0069] Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall;

[0070] The displacement increment is expressed as:

[0071] Δx = SVM(D, P) a ,P max ,P t )

[0072] Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

[0073] Optionally, in one embodiment of this application, the rainfall scenario is determined by rainfall characteristics, and landslide displacement prediction curves under different rainfall scenarios are obtained based on a prediction model, including:

[0074] By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

[0075] Optionally, in one embodiment of this application, the soil mechanical parameters are the rock mechanical parameters of different strata, determined by geotechnical tests, including parameters such as unit weight γ, soil strength parameters (internal cohesion c, internal friction angle φ), elastic modulus, Poisson's ratio, and unit weight. The strength parameters of the landslide body, mainly the slip zone soil, are determined by laboratory direct shear tests. The aforementioned soil mechanical parameters are also determined by inversion from calculated parameters.

[0076] The parameter inversion process includes: roughly determining the range of strength parameters, determining the approximate safety factor based on the current stability of the slope, calculating the strength parameters back using the safety factor, or obtaining a safety factor that is close to the determined value through continuous trial calculation of the strength parameters.

[0077] The landslide range includes the landslide surface displacement, which is determined by on-site geological investigation. The profile is selected based on the current slope stability, with more dangerous or concerned sections as the main profiles. According to the apparent displacement of the landslide, the surface with the largest displacement is selected as the calculation profile for numerical analysis. The two-dimensional numerical analysis methods mainly include the finite element method, the finite difference method, and the discrete element method.

[0078] Specifically, a two-dimensional numerical analysis of the calculated profile is performed based on soil mechanical parameters, including:

[0079] Two-dimensional numerical analysis of the calculated profile was performed using the continuous media mechanics analysis software (FLAC) to obtain the profile displacement and plastic zone;

[0080] The internal cohesion c and internal friction angle φ were gradually reduced proportionally using the strength reduction method. Numerical calculations were performed based on the selected penetration rates of different plastic zones of the slip zone to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

[0081] Taking the plastic zone penetration rates of 85% and 90% of the sliding band as examples, the plastic zone distribution map and the corresponding displacement distribution cloud map are shown below. Figure 3 and Figure 4 , Figure 3 The figures show the distribution of the plastic zone when the penetration rate of the plastic zone of the sliding belt is 85% and 90%, respectively. Figure 4 The figures show the cross-sectional displacement contours when the plastic zone penetration rate of the sliding belt is 85% and 90%, respectively.

[0082] There is a corresponding relationship between the penetration rate of the plastic zone and the displacement cloud map. Based on the plastic zone distribution map and displacement cloud map calculated above, it is determined that when the penetration rate of the plastic zone of the sliding strip reaches 85% and 90%, the displacement values ​​of the corresponding apparent monitoring points (key points) are 454.5 mm and 489.9 mm, respectively.

[0083] Based on the displacement values ​​of 454.5 mm and 489.9 mm at the aforementioned apparent monitoring points, and combined with the landslide displacement prediction curves for the two rainfall scenarios (see...), Figure 2 ),from Figure 2 This allows us to separately determine the rainfall scenarios 1 and 2, and the time it takes for the landslide plastic zone penetration rate to increase from 85% to 90% (see...). Figure 5 When the plastic zone penetration rate reaches 85% and 90%, the corresponding key point displacements are 454.5 mm and 489.9 mm, respectively. The horizontal axis corresponding to the displacement values ​​is the time when the plastic zone of the landslide slide zone evolves to 85% and 90%.

[0084] To achieve the above embodiments, this application also proposes a rainfall-based landslide early warning device based on machine learning models and numerical analysis.

[0085] Figure 6This is a schematic diagram of the structure of a rainfall-induced landslide early warning device based on a machine learning model and numerical analysis, provided in an embodiment of this application.

[0086] like Figure 6 As shown, the rainfall-induced landslide early warning device based on the machine learning model and numerical analysis includes a construction module, a first calculation module, a second calculation module, a third calculation module, and an early warning module.

[0087] The module is used to build a landslide displacement prediction model based on displacement data and rainfall data from apparent landslide monitoring points, and to obtain landslide displacement prediction curves under different rainfall scenarios based on the prediction model.

[0088] The first calculation module is used to determine the landslide range based on geological surveys and to obtain soil mechanical parameters through geotechnical tests and calculation parameter inversion.

[0089] The second calculation module is used to select the calculation profile according to the landslide range, and perform two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain the plastic zone distribution map and displacement distribution cloud map.

[0090] The third calculation module is used to calculate the displacement threshold at the corresponding apparent monitoring point when the plastic zone of the sliding strip evolves to different penetration rates, based on the plastic zone distribution map and displacement cloud map.

[0091] The early warning module is used to determine the time for the displacement to evolve to the threshold displacement based on the displacement threshold of the apparent monitoring points and the landslide displacement prediction curve under different rainfall scenarios, and to implement landslide safety early warning based on the evolution time and early warning level.

[0092] Optionally, in one embodiment of this application, the displacement data is a displacement time series, and the rainfall data is an external rainfall series. The external rainfall series is a sequence composed of rainfall characteristics, including cumulative rainfall, rainfall intensity, and rainfall duration. A landslide displacement prediction model is constructed based on the displacement data and rainfall data from the apparent monitoring points of the landslide, including:

[0093] A training set was constructed based on displacement time series and external rainfall series.

[0094] The external rainfall sequence and displacement time series are used as inputs to the support vector machine model, and the displacement increment is used as the output of the support vector machine model. The support vector machine model is trained using a training set to obtain a landslide displacement prediction model. The kernel parameters of the support vector machine model are globally optimized by the particle swarm optimization algorithm.

[0095] Optionally, in one embodiment of this application, the external rainfall sequence is represented as:

[0096] P = {P} a P max Pt}

[0097] Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall;

[0098] The displacement increment is expressed as:

[0099] Δx = SVM(D, P) a ,P max ,P t )

[0100] Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

[0101] Optionally, in one embodiment of this application, the rainfall scenario is determined by rainfall characteristics, and landslide displacement prediction curves under different rainfall scenarios are obtained based on a prediction model, including:

[0102] By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

[0103] Optionally, in one embodiment of this application, the soil mechanical parameters include internal cohesion and internal friction angle, the landslide range includes landslide surface displacement, and a two-dimensional numerical analysis of the calculated profile is performed based on the soil mechanical parameters, including:

[0104] Two-dimensional numerical analysis of the calculated profile was performed using continuum mechanics analysis software to obtain the profile displacement and plastic zone.

[0105] The internal cohesion and internal friction angle were gradually reduced proportionally using the strength reduction method. Numerical calculations were performed based on the selected penetration rates of different plastic zones of the slip zone to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

[0106] It should be noted that the foregoing explanation of the embodiment of the rainfall-induced landslide early warning method based on machine learning models and numerical analysis also applies to the rainfall-induced landslide early warning device based on machine learning models and numerical analysis in this embodiment, and will not be repeated here.

[0107] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0108] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0109] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0110] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0111] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0112] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it includes one or a combination of the steps of the method embodiments.

[0113] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0114] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A rainfall-based landslide early warning method based on machine learning models and numerical analysis, characterized in that, Includes the following steps: A landslide displacement prediction model is constructed based on displacement data and rainfall data from apparent landslide monitoring points, and landslide displacement prediction curves under different rainfall scenarios are obtained based on the prediction model. The extent of the landslide was determined based on geological surveys, and the soil mechanical parameters were obtained through geotechnical tests. A calculation profile is selected based on the landslide range, and a two-dimensional numerical analysis is performed on the calculation profile based on the soil mechanical parameters to obtain a plastic zone distribution map and a displacement distribution cloud map. Based on the plastic zone distribution map and the displacement cloud map, calculate the displacement threshold at the corresponding apparent monitoring point when the plastic zone of the sliding strip evolves to different penetration rates; Based on the displacement threshold of the apparent monitoring points and the landslide displacement prediction curves under different rainfall scenarios, the time for the displacement to evolve to the threshold displacement is determined, and landslide safety warnings are implemented based on the evolution time and warning level.

2. The method as described in claim 1, characterized in that, The displacement data is a displacement time series, and the rainfall data is an external rainfall series. The external rainfall series is a sequence composed of rainfall characteristics, including cumulative rainfall, rainfall intensity, and rainfall duration. The construction of a landslide displacement prediction model based on the displacement data and rainfall data from landslide apparent monitoring points includes: A training set is constructed based on the displacement time series and the external rainfall series; The external rainfall sequence and displacement time series are used as inputs to the support vector machine model, and the displacement increment is used as the output of the support vector machine model. The support vector machine model is trained using the training set to obtain the landslide displacement prediction model. The kernel parameters of the support vector machine model are globally optimized by the particle swarm optimization algorithm.

3. The method as described in claim 2, characterized in that, The external rainfall sequence is represented as follows: P={P a ,P max ,P t } Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall; The displacement increment is expressed as: ΔxSVM(D,P a ,P max ,P t ) Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

4. The method as described in claim 2, characterized in that, The rainfall scenario is determined by rainfall characteristics, and the process of obtaining landslide displacement prediction curves under different rainfall scenarios based on the prediction model includes: By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

5. The method as described in claim 1, characterized in that, The soil mechanical parameters include internal cohesion and internal friction angle; the landslide range includes landslide surface displacement; and the two-dimensional numerical analysis of the calculated profile based on the soil mechanical parameters includes: Two-dimensional numerical analysis of the calculated profile was performed using continuum mechanics analysis software to obtain the profile displacement and plastic zone. The internal cohesion and internal friction angle are gradually reduced proportionally using the strength reduction method. Numerical calculations are performed based on the selected plastic zone penetration rates of different sliding zones to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

6. A rainfall-based landslide early warning device based on machine learning models and numerical analysis, characterized in that, It includes a construction module, a first calculation module, a second calculation module, a third calculation module, and an early warning module, among which, The construction module is used to construct a landslide displacement prediction model based on displacement data and rainfall data from landslide apparent monitoring points, and to obtain landslide displacement prediction curves under different rainfall scenarios based on the prediction model. The first calculation module is used to determine the landslide range based on geological surveys and to obtain soil mechanical parameters through rock and soil tests; The second calculation module is used to select a calculation profile according to the landslide range, and perform two-dimensional numerical analysis on the calculation profile based on the soil mechanical parameters to obtain a plastic zone distribution map and a displacement distribution cloud map; The third calculation module is used to calculate the displacement threshold at the corresponding apparent monitoring point when the plastic zone of the sliding strip evolves to different penetration rates, based on the plastic zone distribution map and the displacement cloud map. The early warning module is used to determine the time for the displacement to evolve to the threshold displacement based on the displacement threshold of the apparent monitoring point and the landslide displacement prediction curve under different rainfall scenarios, and to implement landslide safety early warning based on the evolution time and early warning level.

7. The apparatus as claimed in claim 6, characterized in that, The displacement data is a displacement time series, and the rainfall data is an external rainfall series. The external rainfall series is a sequence composed of rainfall characteristics, including cumulative rainfall, rainfall intensity, and rainfall duration. The construction of a landslide displacement prediction model based on the displacement data and rainfall data from landslide apparent monitoring points includes: A training set is constructed based on the displacement time series and the external rainfall series; The external rainfall sequence and displacement time series are used as inputs to the support vector machine model, and the displacement increment is used as the output of the support vector machine model. The support vector machine model is trained using the training set to obtain the landslide displacement prediction model. The kernel parameters of the support vector machine model are globally optimized by the particle swarm optimization algorithm.

8. The apparatus as claimed in claim 6, characterized in that, The external precipitation sequence is represented as follows: P={P a ,P max ,P t } Among them, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall; The displacement increment is expressed as: ΔxSVM(D,P a ,P max ,P t ) Where D is the displacement time series, P a P represents the cumulative rainfall. max For rainfall intensity, P t Indicates the duration of rainfall.

9. The apparatus as claimed in claim 6, characterized in that, The rainfall scenario is determined by rainfall characteristics, and the process of obtaining landslide displacement prediction curves under different rainfall scenarios based on the prediction model includes: By inputting the rainfall characteristics corresponding to different rainfall scenarios and the displacement time series of landslide apparent monitoring points into the landslide displacement prediction model, landslide displacement prediction curves of landslide apparent monitoring points under different rainfall scenarios are obtained.

10. The apparatus as claimed in claim 6, characterized in that, The soil mechanical parameters include internal cohesion and internal friction angle; the landslide range includes landslide surface displacement; and the two-dimensional numerical analysis of the calculated profile based on the soil mechanical parameters includes: Two-dimensional numerical analysis of the calculated profile was performed using continuum mechanics analysis software to obtain the profile displacement and plastic zone. The internal cohesion and internal friction angle are gradually reduced proportionally using the strength reduction method. Numerical calculations are performed based on the selected plastic zone penetration rates of different sliding zones to obtain plastic zone distribution maps and corresponding displacement distribution cloud maps.

Images