Multi-scale earthquake landslide risk evaluation method based on artificial intelligence

Abstract

The invention relates to the technical field of geological disaster prediction and evaluation, and particularly discloses a multi-scale earthquake landslide risk evaluation method based on artificial intelligence, and the method comprises the steps: firstly collecting and sorting the earthquake oscillation time history, high-resolution terrain, rock-soil body parameters and monitoring data of a single slope scale; seismic oscillation intensity parameter zoning of the regional slope scale, a digital elevation model with medium and low resolution, a geological structure and the like are obtained; establishing a regional scale earthquake landslide risk prediction model based on a gradient lifting decision tree algorithm to obtain a regional earthquake landslide risk probability; for the high-risk area, an improved Newmark sliding block method is adopted to carry out single slope instability analysis, and the slope instability probability is obtained; through regional scale risk evaluation and monomer slope stability evaluation, a'monomer-region 'hierarchical integration mechanism and a'region-monomer' hierarchical integration mechanism are constructed respectively, and monomer-region cross-scale earthquake slope risk dynamic evaluation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster prediction and evaluation, and in particular to a multi-scale earthquake landslide hazard evaluation method based on artificial intelligence. Background Art

[0002] Landslide is one of the main forms of earthquake-induced geological disasters. Especially under the action of strong earthquakes, the distribution and scale of landslide disasters in complex mountainous terrain are highly uncertain.

[0003] Scholars have conducted a lot of research at the regional scale. Scholars from institutions such as the China Earthquake Administration, the National Institute of Disaster Prevention and Control of the Ministry of Emergency Management, the United States Geological Survey, Southwest Jiaotong University, and Chengdu University of Technology have proposed a number of slope hazard assessment models based on physical mechanics and data analysis methods. In 2000, Jibson et al. (2000) evaluated the slope hazard impact area of ​​the 1994 Northridge earthquake in the United States in combination with GIS, and proposed the use of Weibull function to analyze the relationship between permanent displacement and slope hazard probability. With the development of artificial intelligence technology, machine learning models have begun to show significant advantages in landslide hazard prediction, especially the ability to process large-scale data. However, these data-driven models usually lack an understanding of physical mechanisms, and the prediction results are not interpretable enough, which affects their credibility and universality in practical applications.

[0004] In the existing related technical research, the inventors found that the earthquake landslide evaluation model is mainly based on the analysis of a single scale, lacking the coupling analysis of the regional scale and the single slope scale, and it is difficult to effectively reflect the comprehensive effect of multi-source data such as earthquake source, topography, and rock and soil characteristics. The traditional model still has shortcomings in multi-scale integration, and the refined research of a single scale does not match the wide-area research of the regional scale. Therefore, the present invention aims to solve the problems of inefficiency and limitations in the existing technology for predicting landslides in mountainous areas with strong earthquakes, and proposes a multi-scale earthquake landslide hazard evaluation method based on artificial intelligence. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention provides a multi-scale earthquake landslide hazard assessment method based on artificial intelligence. It is a multi-scale earthquake landslide hazard intelligent assessment method that integrates a single-scale model and a regional scale model. It is particularly suitable for the dynamic assessment of slope disaster risks in mountainous areas with strong earthquakes, and solves the problems mentioned in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a multi-scale earthquake landslide hazard assessment method based on artificial intelligence, comprising the following steps:

[0007] S1, multi-scale data collection and data processing;

[0008] S2. Regional scale earthquake landslide hazard assessment;

[0009] S3, single-scale seismic slope stability analysis;

[0010] S4. Construct a multi-scale evaluation framework and dynamic correction.

[0011] Preferably, in step S1, historical earthquake landslide data from all over the world are collected, and data including seismic data, topography, geology and deformation are collected from the single-body scale and the regional scale according to the spatial location of the landslide.

[0012] Preferably, at the monomer scale, it specifically includes the following:

[0013] A. Collect historical earthquake time history data, including earthquake acceleration time history, velocity time history, and displacement time history; calculate energy-based earthquake intensity parameters including Arias intensity and pulse energy rate based on earthquake time history;

[0014] B. Collect high-resolution terrain data through drone aerial photography or lidar, including slope geometry parameters such as slope gradient, slope aspect, and slope height;

[0015] C. Collect geological data and obtain the mechanical parameters of the rock and soil mass of the slope, including cohesion, internal friction angle and gravity;

[0016] D. Obtain dynamic change information of stress, displacement or inclination through existing on-site slope monitoring data.

[0017] Preferably, at the regional scale, it specifically includes the following:

[0018] 1) Perform spatial interpolation analysis through GIS to obtain the seismic zoning map of the study area, including peak acceleration and peak velocity;

[0019] 2) Based on the national geospatial database and NASA platform, obtain low- and medium-resolution terrain data, including regional slope, aspect, and height data;

[0020] 3) Obtain regional geological structure information, including fault distribution and lithology grouping;

[0021] 4) Obtain satellite remote sensing data, including data on ground subsidence and surface deformation.

[0022] Preferably, in step S2, the following is specifically included:

[0023] S21. Extract various types of regional-scale seismic data, topography, geology and deformation data as influencing factors, and perform [0,1] standardization on all data; obtain the correlation values ​​between factors through bivariate correlation analysis, and retain factors with correlations with other influencing factors less than 0.8 as input eigenvalue variables;

[0024] Whether a landslide occurs is taken as the target value, with 1 being the occurrence of a landslide and 0 being the non-occurrence of a landslide; the characteristic value variables and the target values ​​are aggregated to form a data set;

[0025] S22, based on the gradient boosting decision tree algorithm, establish an earthquake landslide hazard assessment model, first select the logarithmic loss function as the algorithm objective function, set the model's hyperparameters, including learning rate, maximum depth, and number of numbers; divide the data set of step S21 into a training set, a validation set, and a test set according to a 6:2:2 ratio, train the model with the test set data, evaluate the model performance with the validation set, adjust the hyperparameters, and calculate the model's evaluation indicators based on the test set, including accuracy, AUC value, and precision; finally output the landslide hazard probability P region .

[0026] Preferably, in step S22, according to the risk probability P region The danger level is divided into two categories: the extremely high danger zone is defined as the danger probability greater than 80%, the high danger zone is defined as 60%-80%, the medium danger zone is defined as 40%-60%, the low danger zone is defined as 20%-40%, and the extremely low danger zone is defined as the danger probability less than 20%.

[0027] Preferably, in step S3, various types of ground motion data, topography, and geological data at a monomer scale are extracted, and the permanent displacement of the slope is calculated based on the improved dynamic slider method in combination with the ground motion time history, the geometric characteristics of the slope, and the mechanical parameters of the rock and soil body. The calculation formula is as follows:

[0028]

[0029] Among them, D N is the permanent displacement of the slope; a(t) is the earthquake acceleration time history; a c is the initial critical acceleration of the slope; t is the time;

[0030] Based on the relationship between the permanent displacement of the slope and the probability of slope instability established by the Weibull function, the slope instability probability P is calculated. instable , and its calculation formula is as follows:

[0031]

[0032] Preferably, in step S4, through regional-scale earthquake landslide hazard assessment and single-scale earthquake slope stability analysis, a multi-scale evaluation framework of "single-region" and "region-single-region" is constructed, and dynamic correction of the evaluation results is achieved.

[0033] Preferably, in step S4, the following steps are specifically included:

[0034] S41. “Single-region” cross-scale evaluation:

[0035] Through the regional-scale earthquake landslide hazard evaluation in step S2, an extremely high-hazard area is obtained, and the number of extremely high-hazard slopes in the area is identified as individual slopes;

[0036] Obtain these single-scale characteristic values, including earthquake acceleration time history, velocity time history, slope and rock and soil parameters, and then bring them into the single-scale earthquake slope stability analysis in step S3 to obtain the instability probability of these high-risk single-scale slopes;

[0037] S42. In a certain study area, characteristic values ​​of individual single slopes can be obtained, including earthquake acceleration time history, velocity time history, slope, and rock and soil parameters. Based on the single-scale earthquake slope stability analysis method in step S3, the instability probability of these single slopes is obtained, and the probability value is normalized;

[0038] Then, the earthquake landslide hazard probability of the study area is obtained through step S2, and the landslide hazard probability corresponding to the space in the area is corrected by the normalized instability probability of the single slope, so that the corrected regional earthquake landslide hazard probability value can be obtained. The calculation formula is as follows:

[0039]

[0040] in, is the corrected earthquake landslide hazard probability; α is the correction intensity, ranging from 0 to 1, 1 means directly replacing the hazard probability of the corresponding slope in the area with the instability probability of the single slope, and 0 means no correction.

[0041] Preferably, in step S42, the normalization means that when P instable When =0.335, the normalized probability value is 1.

[0042] The beneficial effects of the present invention are as follows: the present invention combines the monomer scale model with the regional scale model, and through regional scale hazard evaluation and monomer slope stability evaluation, respectively constructs a hierarchical integration mechanism of "monomer-region" and "region-monomer", thereby realizing a dynamic evaluation of monomer-region cross-scale seismic slope hazard. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1Schematic diagram of the process of a multi-scale earthquake landslide hazard assessment method based on artificial intelligence in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present invention provides a technical solution: a multi-scale earthquake landslide hazard assessment method based on artificial intelligence, such as Figure 1 As shown, the following steps are included:

[0046] S1. Multi-scale data collection and data processing.

[0047] In step S1, historical earthquake landslide data from all over the world are collected. According to the spatial location of the landslide, ground motion data, topography, geology and deformation data are collected from the single scale and regional scale respectively. As shown in Table 1:

[0048] Table 1 Data types at the monomer scale and regional scale

[0049] Data Types Single slope scale Regional slope scale Earthquake data Earthquake time history data Earthquake zoning map Terrain data High-resolution terrain data Medium and low resolution terrain data Geological data Mechanical parameters of rock and soil Regional geological structure map Deformation data Monitoring data Remote sensing data

[0050] There are two ways to obtain data:

[0051] At the monomer scale, it includes the following:

[0052] A. Collect historical earthquake time history data from channels such as the China Earthquake Networks Center and the Pacific Earthquake Engineering Research Center, including earthquake acceleration time history, velocity time history, and displacement time history; calculate energy-based earthquake intensity parameters including Arias intensity and pulse energy rate based on earthquake time history;

[0053] B. Collect high-resolution terrain data through drone aerial photography or laser radar (LiDAR), including slope geometry parameters such as slope gradient, slope aspect, and slope height;

[0054] C. Collect geological data based on literature and obtain the mechanical parameters of the rock and soil mass of the slope, including cohesion, internal friction angle and gravity;

[0055] D. Obtain dynamic change information of stress, displacement or inclination through existing on-site slope monitoring data.

[0056] At the regional scale, these include:

[0057] 1) Based on the seismic monitoring stations, spatial interpolation analysis is performed through GIS to obtain the seismic zoning map of the study area, including peak acceleration (PGA), peak velocity (PGV), etc.;

[0058] 2) Based on the national geospatial database, NASA and other platforms, obtain low- and medium-resolution terrain data: such as digital elevation models (DEMs), and process DEM data based on GIS software, including regional slope, aspect, and height data;

[0059] 3) Regional geological structure information, including fault distribution and lithology grouping, can be obtained through the National Geological Data Center and One Geology Global platform;

[0060] 4) Obtain satellite remote sensing data such as Landsat and Sentinel through public platforms such as USGS Earth Explorer and ESA Copernicus Open Access Hub; perform data processing based on ENVI (Environment for Visualizing Images) software to obtain data such as ground subsidence and surface deformation.

[0061] The single data is time-aligned based on the earthquake time; the regional data is spatially aligned based on the longitude and latitude coordinates using GIS to ensure that all data are in the same projection coordinate system and have the same spatial resolution.

[0062] S2. Regional-scale earthquake landslide hazard assessment.

[0063] In step S2, the specific steps include:

[0064] S21, based on the historical earthquake landslide data collected in step S1, extract various types of regional-scale seismic data, topography, geology and deformation data as influencing factors, and perform [0,1] standardization on all data; obtain the correlation values ​​between factors through bivariate correlation analysis, and retain factors with a correlation with other influencing factors less than 0.8 as input eigenvalue variables;

[0065] Whether a landslide occurs is taken as the target value, with 1 being the occurrence of a landslide and 0 being the non-occurrence of a landslide; the characteristic value variables and the target values ​​are aggregated to form a data set;

[0066] S22, based on the gradient boosting decision tree (GBDT) algorithm, establish an earthquake landslide hazard assessment model, first select the logarithmic loss function (Logarithmic Loss) as the algorithm objective function, set the model's hyperparameters, including learning rate, maximum depth, and number of numbers; divide the data set in step S21 into a training set, a validation set, and a test set according to a 6:2:2 ratio, train the model with the test set data, evaluate the model performance with the validation set, adjust the hyperparameters, and calculate the model's evaluation indicators based on the test set, including accuracy, AUC value, and precision; finally output the landslide hazard probability P region .

[0067] According to the risk probability P region The danger level is divided into two categories: the extremely high danger zone is defined as the danger probability greater than 80%, the high danger zone is defined as 60%-80%, the medium danger zone is defined as 40%-60%, the low danger zone is defined as 20%-40%, and the extremely low danger zone is defined as the danger probability less than 20%.

[0068] S3. Single-scale earthquake slope stability analysis.

[0069] In step S3, various types of ground motion data, topography, and geological data at the monomer scale are extracted, including ground motion time history, detailed slope geometric characteristics (slope, slope aspect, slope height, etc.) and rock and soil mechanical parameters (cohesion, internal friction angle, gravity, etc.); based on the improved dynamic slider method, the permanent displacement of the slope is calculated by combining the ground motion time history, slope geometric characteristics, and rock and soil mechanical parameters. The calculation formula is as follows:

[0070]

[0071] Among them, D N is the permanent displacement of the slope, in cm; a(t) is the earthquake acceleration time history, in g; a c is the initial critical acceleration of the slope, in g; t is the time, in s;

[0072] Based on the relationship between the permanent displacement of the slope and the probability of slope instability established by the Weibull function, the slope instability probability P is calculated. instable , and its calculation formula is as follows:

[0073]

[0074] Where 0.335 is the maximum value of the slope instability probability, -0.048 and 1.565 are coefficients; this value may change depending on the historical landslide data used.

[0075] S4. Construct a multi-scale evaluation framework and dynamic correction.

[0076] The requirements for data refinement at the monomer scale and regional scale are different, see Table 2:

[0077] Table 2 Comparison of data at the monomer scale and regional scale

[0078] Single slope scale Regional slope scale Earthquake acceleration time history Peak acceleration Earthquake velocity time history Peak speed Slope, slope direction (fine) Slope, aspect (10m resolution) Mechanical parameters of rock and soil Lithology Group Probability of slope failure Hazard level

[0079] The calculation of single slopes is more sophisticated and complex, and its evaluation results are more accurate. However, if a single-scale earthquake slope stability analysis is to be performed on all slopes in a large area, it will take a lot of time and computing resources. On the contrary, for the regional scale, its data acquisition is easier and rougher, and the evaluation results are less accurate than those of single slopes, but its efficiency is higher. Therefore, in order to ensure both calculation accuracy and efficiency. Through regional-scale earthquake landslide hazard evaluation and single-scale earthquake slope stability analysis, a multi-scale evaluation framework of "single-region" and "region-single" is constructed, and dynamic correction of evaluation results is achieved. Specifically, it includes the following:

[0080] S41. “Single-region” cross-scale evaluation:

[0081] Through the regional-scale earthquake landslide hazard assessment in step S2, an extremely high-hazard area (hazard probability greater than 80%) is obtained, and the number of extremely high-hazard slopes in the area is identified as individual slopes;

[0082] Obtain these single-scale characteristic values, including earthquake acceleration time history, velocity time history, slope and rock and soil parameters, and then bring them into the single-scale seismic slope stability analysis in step S3 to obtain the instability probability of these highly dangerous single slopes. Through the real-time update of remote sensing data, seismic load data, etc. at the regional scale, the regional scale seismic slope hazard map can be updated at the same time, thereby completing the refined dynamic evaluation of single slopes.

[0083] S42. In a certain study area, the characteristic values ​​of individual single slopes can be obtained, including earthquake acceleration time history, velocity time history, slope, and rock and soil parameters. Based on the single-scale earthquake slope stability analysis method in step S3, the instability probability of these single slopes is obtained, and the probability value is normalized (when P instable =0.335, the normalized probability value is 1);

[0084] Then, the earthquake landslide hazard probability of the study area is obtained through step S2, and the landslide hazard probability corresponding to the space in the area is corrected by the normalized instability probability of the single slope, so that the corrected regional earthquake landslide hazard probability value can be obtained. The calculation formula is as follows:

[0085]

[0086] in, is the corrected earthquake landslide hazard probability; α is the correction intensity, ranging from 0 to 1, 1 means directly replacing the hazard probability of the corresponding slope in the area with the instability probability of the single slope, and 0 means no correction.

[0087] The present invention combines the single-scale model with the regional-scale model, and through regional-scale hazard evaluation and single-scale slope stability evaluation, constructs a hierarchical integration mechanism of "single-region" and "region-single-region" respectively, to achieve a dynamic evaluation of single-region cross-scale seismic slope hazard.

[0088] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0089] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0090] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0091] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0092] The "first\second" mentioned in the embodiments is only to distinguish similar objects, and does not represent a specific order for the objects. It is understandable that the "first\second" can be interchanged with the specific order or sequence where permitted. It should be understood that the objects distinguished by "first\second" can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than those illustrated or described herein.

[0093] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-scale earthquake landslide hazard assessment method based on artificial intelligence, characterized in that: The steps include: S1, multi-scale data collection and data processing; S2. Regional scale earthquake landslide hazard assessment; S3, single-scale seismic slope stability analysis; S4. Construct a multi-scale evaluation framework and dynamic correction.

2. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 1 is characterized by: In step S1, historical earthquake landslide data from all over the world are collected, and data including seismic data, topography, geology and deformation are collected from the single-unit scale and the regional scale according to the spatial location of the landslide.

3. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 2 is characterized by: At the monomer scale, it includes the following: A. Collect historical earthquake time history data, including earthquake acceleration time history, velocity time history, and displacement time history; calculate energy-based earthquake intensity parameters including Arias intensity and pulse energy rate based on earthquake time history; B. Collect high-resolution terrain data through drone aerial photography or lidar, including slope geometry parameters such as slope gradient, slope aspect, and slope height; C. Collect geological data and obtain the mechanical parameters of the rock and soil mass of the slope, including cohesion, internal friction angle and gravity; D. Obtain dynamic change information of stress, displacement or inclination through existing on-site slope monitoring data.

4. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 2 is characterized by: At the regional scale, these include: 1) Perform spatial interpolation analysis through GIS to obtain the seismic zoning map of the study area, including peak acceleration and peak velocity; 2) Based on the national geospatial database and NASA platform, obtain low- and medium-resolution terrain data, including regional slope, aspect, and height data; 3) Obtain regional geological structure information, including fault distribution and lithology grouping; 4) Obtain satellite remote sensing data, including data on ground subsidence and surface deformation.

5. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 1 is characterized by: In step S2, the specific steps include: S21, extracting various types of regional-scale seismic data, topography, geology and deformation data as influencing factors, and performing [0,1] standardization on all data; Through bivariate correlation analysis, the correlation values ​​between factors are obtained, and the factors with correlations with other influencing factors less than 0.8 are retained as input eigenvalue variables; Whether a landslide occurs is taken as the target value, with 1 being the occurrence of a landslide and 0 being the non-occurrence of a landslide; the characteristic value variables and the target values ​​are aggregated to form a data set; S22. Based on the gradient boosting decision tree algorithm, an earthquake landslide hazard assessment model is established. First, the logarithmic loss function is selected as the algorithm objective function, and the hyperparameters of the model are set, including the learning rate, the maximum depth, and the number of numbers; The data set in step S21 is divided into a training set, a validation set, and a test set according to a ratio of 6:2:

2. The test set data is used for model training, the validation set is used to evaluate the model performance, and the hyperparameters are adjusted. The evaluation indicators of the model are calculated based on the test set, including accuracy, AUC value, and precision rate. Finally, the landslide hazard probability P is output. region .

6. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 5 is characterized by: In step S22, according to the risk probability P region The danger level is divided into two categories: the extremely high danger zone is defined as the danger probability greater than 80%, the high danger zone is defined as 60%-80%, the medium danger zone is defined as 40%-60%, the low danger zone is defined as 20%-40%, and the extremely low danger zone is defined as the danger probability less than 20%.

7. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 1 is characterized by: In step S3, various types of ground motion data, topography, and geological data at the monomer scale are extracted, and the permanent displacement of the slope is calculated based on the improved dynamic slider method, combined with the ground motion time history, slope geometric characteristics, and rock and soil mechanical parameters. The calculation formula is as follows: Among them, D N is the permanent displacement of the slope; a(t) is the earthquake acceleration time history; a c is the initial critical acceleration of the slope; t is the time; Based on the relationship between the permanent displacement of the slope and the probability of slope instability established by the Weibull function, the slope instability probability P is calculated. instable , and its calculation formula is as follows:

8. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 1 is characterized by: In step S4, through regional-scale earthquake landslide hazard assessment and single-scale earthquake slope stability analysis, a multi-scale assessment framework of "single-region" and "region-single-region" is constructed, and the dynamic correction of the assessment results is realized.

9. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 8 is characterized by: In step S4, the following steps are specifically included: S41, "Single-region" cross-scale evaluation: Through the regional-scale earthquake landslide hazard evaluation in step S2, an extremely high-hazard area is obtained, and the number of extremely high-hazard slopes in the area is identified as individual slopes; Obtain these single-scale characteristic values, including earthquake acceleration time history, velocity time history, slope and rock and soil parameters, and then bring them into the single-scale earthquake slope stability analysis in step S3 to obtain the instability probability of these high-risk single-scale slopes; S42. In a certain study area, characteristic values ​​of individual single slopes can be obtained, including earthquake acceleration time history, velocity time history, slope, and rock and soil parameters. Based on the single-scale earthquake slope stability analysis method in step S3, the instability probability of these single slopes is obtained, and the probability value is normalized; Then, the earthquake landslide hazard probability of the study area is obtained through step S2, and the landslide hazard probability corresponding to the space in the area is corrected by the normalized instability probability of the single slope, so that the corrected regional earthquake landslide hazard probability value can be obtained. The calculation formula is as follows: in, is the corrected earthquake landslide hazard probability; α is the correction intensity, ranging from 0 to 1, 1 means directly replacing the hazard probability of the corresponding slope in the area with the instability probability of the single slope, and 0 means no correction.

10. The multi-scale earthquake landslide hazard assessment method based on artificial intelligence according to claim 9 is characterized by: In step S42, the normalization means that when P instable When =0.335, the normalized probability value is 1.

Images