A method and system for geological hazard assessment based on machine learning

Abstract

The present invention proposes a method and system for evaluating the risk of geological disasters based on machine learning. It belongs to the technical field of machine learning and geological disaster assessment; the method includes: determining the target area of ​​geological disasters to be evaluated, obtaining multi-source data of the target area, processing the original data based on ArcGIS, dividing the grid units, and extracting the geological disaster related factor data of each grid unit; selecting geological disaster susceptibility evaluation indicators according to the mechanism of geological disaster occurrence and regional characteristics; and determining geological disaster vulnerability evaluation indicators; performing positive and normalization processing on the selected evaluation indicators, and constructing a geological disaster risk evaluation indicator system. By adopting a variety of machine learning algorithms and parameter optimization techniques, the risk of geological disasters can be predicted more accurately, thereby reducing the false alarm and missed alarm rates.

Description

Technical Field

[0001] The present invention proposes a geological hazard risk assessment method and system based on machine learning, which belongs to the technical field of machine learning and geological hazard assessment. Background Art

[0002] With the global climate change and the intensification of human activities, the frequency and complexity of geological disasters have become increasingly prominent, and traditional geological disaster prediction and prevention methods can no longer meet the current needs. Therefore, how to scientifically and accurately assess the risk of geological disasters and formulate corresponding prevention and control strategies has become an important topic in the field of geological disaster research.

[0003] In recent years, with the rapid development of remote sensing technology, GIS (geographic information system) and machine learning technology, new ideas and methods have been provided for geological hazard risk assessment. Remote sensing technology can quickly obtain large-scale, high-precision surface information, providing a rich data source for geological hazard risk assessment; GIS technology has a powerful spatial analysis function, which can integrate, process and analyze multi-source data to reveal the complex relationship between geological hazards and the geographical environment; and machine learning technology, especially deep learning algorithms, can automatically learn rules from data, build prediction models, and achieve accurate prediction of geological hazard risks.

[0004] However, the current geological hazard assessment method still has some shortcomings, such as single data source, incomplete selection of evaluation indicators, and the need to improve the accuracy of model prediction. Therefore, there is an urgent need for a geological hazard assessment method that integrates multi-source data, comprehensively considers the mechanism of geological disasters and regional characteristics, and uses advanced machine learning technology for modeling and prediction. Summary of the invention

[0005] The present invention provides a method and system for evaluating geological hazards based on machine learning to solve the problems mentioned in the above background technology:

[0006] The present invention proposes a method for assessing geological hazards based on machine learning, the method comprising:

[0007] S1. Determine the target area of ​​geological hazards to be evaluated, obtain multi-source data of the target area, process the original data based on ArcGIS, divide the grid units, and extract the geological hazard related factor data of each grid unit;

[0008] S2. Select geological disaster susceptibility evaluation indicators based on the occurrence mechanism and regional characteristics of geological disasters; determine geological disaster vulnerability evaluation indicators; perform positive and normalization processing on the selected evaluation indicators to construct a geological disaster risk evaluation indicator system;

[0009] S3, select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain training sets and test sets;

[0010] S4. Modeling is performed through a variety of machine learning algorithms. The whale algorithm is used to optimize the parameters of the support vector machine model. The training set is input into the optimized machine learning model for training. The trained model is verified using the test set to evaluate the model performance.

[0011] S5. Input the data of the area to be evaluated into the trained geological hazard risk assessment model to make a risk prediction; based on the prediction results, formulate a geological hazard risk classification standard and draw a geological hazard risk distribution map; based on the actual situation in the region, formulate geological hazard prevention and control strategies and emergency plans.

[0012] Furthermore, the S1 includes:

[0013] S11. Determine the specific objectives of geological hazard assessment, and determine the required multi-source data types and accuracy requirements based on historical hazard records, geological exploration reports and expert consultation;

[0014] S12. Collect original remote sensing images through various methods, and use GIS software to perform geometric correction, radiation correction and atmospheric correction on remote sensing images;

[0015] S13. Obtain topographic data and geological structure data from the geological department and perform data consistency verification, and obtain long-term and short-term meteorological data from the meteorological department;

[0016] S14. Use data fusion technology to align multi-source data in space and time;

[0017] S15. Based on the evaluation requirements and data resolution, an adaptive grid division method is used to divide the target area into grid cells of varying sizes but interconnected;

[0018] S16. For each grid unit, use GIS spatial analysis tools to extract geological hazard related factor data.

[0019] Furthermore, the S14 includes:

[0020] Standardize the format, coordinate system and projection of data from different sources, and use statistical methods and image processing techniques to remove noise and outliers in the data;

[0021] Using SIFT and SURF algorithms, stable feature points are extracted from remote sensing images, topographic maps, and geological maps, and accurate matching is performed to establish spatial correspondence between different data sources;

[0022] Based on the feature point matching results, the affine transformation model is used to adjust the relative position of the data set, and the gaps caused by the transformation are filled by interpolation method;

[0023] For time series data, a complete time series framework is constructed to correspond the data at each time point to the evaluation system; time interpolation is used to adjust the time series of each data source to a unified time resolution;

[0024] According to the reliability, resolution and timeliness of the data source, different weights are assigned to each data type, and data fusion is performed through weighted averaging.

[0025] Furthermore, the S15 includes:

[0026] Based on the overall size of the evaluation area and the data resolution, an initial benchmark grid size is set; using GIS software, the evaluation area is preliminarily evenly divided according to the benchmark grid size to generate a preliminary grid framework;

[0027] Analyze the terrain complexity of the evaluation area using topographic and geomorphic data, and identify high-risk and low-risk areas for geological hazards by combining historical disaster records, geological exploration reports and expert knowledge base;

[0028] In high-risk areas or areas with complex terrain, the grid size is reduced according to actual needs to achieve a more refined division; in low-risk areas or areas with flat terrain, the grid size is increased;

[0029] Check whether the data resolution in each grid unit meets the evaluation requirements; for areas with insufficient resolution, further collect high-resolution data or use data interpolation methods to improve data quality;

[0030] GIS spatial analysis tools are used to evaluate the quality of the divided grids, assign a unique code or identifier to each grid unit, and establish an attribute database for the grid units to record the key data of each grid.

[0031] Furthermore, the key data include basic information, terrain features and geological disaster risks.

[0032] Furthermore, the S2 includes:

[0033] S21. Conduct in-depth analysis of geological disaster susceptibility; based on literature research and expert experience, preliminarily screen out a series of potential susceptibility evaluation indicators;

[0034] S22. Use correlation analysis and principal component analysis to reduce the dimension and rank the importance of the indicators initially screened to determine the key evaluation indicators;

[0035] S23. Analyze the density of human activities and potential economic losses in the region in combination with socio-economic data; formulate differentiated vulnerability assessment indicators;

[0036] S24. Based on the analytic hierarchy process, a geological hazard risk assessment index system is constructed. The evaluation indicators are scored and weighted based on the expert knowledge base, and the weight settings are optimized through multiple rounds of iterations.

[0037] Furthermore, the S3 includes:

[0038] S31. Based on the temporal distribution, spatial distribution and type distribution of disaster points, representative disaster points are selected from the historical geological disaster database as positive samples;

[0039] S32, using unsupervised learning methods to generate potential negative sample points outside the buffer zone; combining expert knowledge base and GIS spatial analysis to verify and screen the generated negative sample points;

[0040] S33, based on the geological hazard evaluation index system, extract the evaluation index value of each sample point, and use data enhancement technology to expand the sample set;

[0041] S34. Use stratified sampling or random sampling methods to divide the sample set into a training set and a test set; and further perform data cleaning and preprocessing on the training set and the test set.

[0042] Furthermore, the S4 includes:

[0043] S41. Evaluate the applicability, advantages and disadvantages of different machine learning algorithms in geological hazard assessment, and combine multiple algorithms based on existing ensemble learning schemes;

[0044] S42. Design parameter optimization schemes for each machine learning algorithm, and use meta-heuristic optimization algorithms to globally search and optimize model parameters;

[0045] S43, evaluating the model performance under different parameter combinations through cross-validation, and selecting the optimal parameter combination; inputting the training set into the optimized machine learning model for training;

[0046] S44. Monitor the performance indicators of the model in real time during the training process; and optimize the model according to the training results;

[0047] S45. Use the test set to verify the trained model, draw the ROC curve and PR curve, and analyze the classification performance and generalization ability of the model.

[0048] S46. Interpret and evaluate the model prediction results based on expert knowledge and actual geological hazards.

[0049] Furthermore, the S5 includes:

[0050] S51, inputting the data of the area to be evaluated into the trained geological hazard risk assessment model for prediction, outputting the geological hazard risk score or level of each grid unit, and performing spatial visualization display;

[0051] S52. Combine GIS spatial analysis functions to identify high-risk areas and potential disaster points; formulate geological disaster risk level classification standards based on prediction results;

[0052] S53. Use GIS software to draw a geological hazard risk distribution map to intuitively display the spatial distribution characteristics of regional geological hazard risks;

[0053] S54. Develop specific prevention and control strategies for high-risk areas and potential disaster sites, consider the cost-effectiveness and feasibility of different prevention and control strategies, and develop a comprehensive prevention and control plan;

[0054] S55. Develop emergency response plans for geological disasters, organize emergency drills and training activities, and revise and improve the emergency response plans based on drill results and actual conditions.

[0055] The present invention proposes a geological disaster risk assessment system based on machine learning, the system comprising:

[0056] Data acquisition module: determine the target area of ​​geological hazards to be evaluated, obtain multi-source data of the target area, process the original data based on ArcGIS, divide the grid units, and extract the geological hazard related factor data of each grid unit;

[0057] System construction module: According to the occurrence mechanism and regional characteristics of geological disasters, select geological disaster susceptibility evaluation indicators; determine geological disaster vulnerability evaluation indicators; perform positive and normalization processing on the selected evaluation indicators to construct a geological disaster risk evaluation indicator system;

[0058] Data processing module: select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain the training set and test set;

[0059] Performance evaluation module: Modeling is performed through a variety of machine learning algorithms, the whale algorithm is used to optimize the parameters of the support vector machine model, the training set is input into the optimized machine learning model for training, the trained model is verified using the test set, and the model performance is evaluated;

[0060] Plan formulation module: input the data of the area to be evaluated into the trained geological hazard risk assessment model to make a risk prediction; based on the prediction results, formulate geological hazard risk grade classification standards and draw a geological hazard risk distribution map; based on the actual situation in the region, formulate geological hazard prevention and control strategies and emergency plans.

[0061] The invention has the following beneficial effects: by adopting a variety of machine learning algorithms and parameter optimization techniques, the danger of geological disasters can be predicted more accurately, thereby reducing the false alarm and missed alarm rates; the method takes into account multi-source data, including remote sensing images, topography, geological structure and meteorological data, etc., and can comprehensively evaluate the danger of geological disasters; by grid division and terrain complexity analysis, a refined evaluation of high-risk areas of geological disasters is achieved, which is helpful to formulate targeted prevention and control measures; the grid size is dynamically adjusted according to the terrain complexity and risk level, ensuring the data resolution and evaluation accuracy in different areas; by combining expert knowledge and machine learning technology, a scientific and reasonable evaluation index system is constructed, and socio-economic data and vulnerability analysis are considered at the same time, so that the evaluation results are more practical; by using data fusion technology and GIS spatial analysis tools, the efficiency of data processing and analysis is improved, and the evaluation cycle is shortened; the method provides detailed operation steps, including data collection, processing, modeling, verification and result application, which is convenient for practical operation and promotion; not only the danger evaluation of geological disasters is provided, but also prevention and control strategies and emergency plans are formulated in combination with the evaluation results, which is helpful to improve the disaster prevention and mitigation capabilities; the geological disaster danger distribution map is drawn by GIS software, so that the evaluation results are intuitive and easy to understand, which is convenient for decision makers and the public to understand. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a step diagram of the method of the present invention;

[0063] Figure 2 This is a system module diagram of the present invention. DETAILED DESCRIPTION

[0064] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0065] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than 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.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0067] One embodiment of the present invention, as Figure 1 As shown, a method for assessing geological hazards based on machine learning comprises:

[0068] S1. Determine the target area of ​​geological hazards to be evaluated, obtain multi-source data of the target area (original remote sensing images, topographic data, geological structure data, meteorological data), process the original data based on ArcGIS, divide the grid units, and extract the geological hazard related factor data of each grid unit;

[0069] S2. According to the occurrence mechanism of geological disasters and regional characteristics, select geological disaster susceptibility evaluation indicators (slope, slope direction, slope height, river density, elevation, rock and soil type, distance from faults, distance from rivers, vegetation coverage, rainfall); and determine geological disaster vulnerability evaluation indicators (population density and economic losses); positively and normalize the selected evaluation indicators to construct a geological disaster risk evaluation indicator system;

[0070] S3, select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain training sets and test sets;

[0071] S4. Modeling is performed through a variety of machine learning algorithms (such as support vector machines, random forests, deep learning, etc.), and optimization algorithms such as the whale algorithm are used to optimize the parameters of the support vector machine model; the training set is input into the optimized machine learning model for training to obtain a trained geological hazard assessment model; the trained model is verified using the test set to evaluate the model performance;

[0072] S5. Input the data of the area to be evaluated into the trained geological hazard risk assessment model to make a risk prediction; based on the prediction results, formulate a geological hazard risk classification standard and draw a geological hazard risk distribution map; based on the actual situation in the region, formulate targeted geological hazard prevention and control strategies and emergency plans.

[0073] The working principle of the above technical solution is: clarify the target area of ​​geological disasters to be evaluated, and widely collect various types of data in the area, including original remote sensing images, topographic data, geological structure data, and meteorological data. These data are the basis for subsequent analysis. Through ArcGIS and other geographic information system software, these original data are processed, such as divided into grid units, to facilitate subsequent data analysis and feature extraction. In each grid unit, extract factor data related to geological disasters, such as slope, slope aspect, elevation, etc.; based on the occurrence mechanism and regional characteristics of geological disasters, select a series of evaluation indicators, which include both geological disaster susceptibility evaluation indicators (such as slope, slope aspect, river density, rock and soil type, etc.) and vulnerability evaluation indicators (such as population density, economic losses, etc.). The selected evaluation indicators are forward-oriented and normalized to ensure that indicators of different dimensions and magnitudes can be compared and calculated in the same evaluation system, so as to construct a complete geological disaster risk evaluation indicator system; in order to train the machine learning model, a sample set needs to be prepared. Historical geological disaster points are usually selected as positive samples, and buffer zones are set around these points to ensure that the model can learn the characteristics of these disaster points. Negative samples are randomly selected outside the buffer zone. The selection of negative samples should ensure the rationality of the spatial range to simulate the situation in non-disaster areas. Based on the geological disaster risk evaluation index system, the sample set is preprocessed, including feature extraction, data cleaning, etc., and finally the training set and test set are obtained. A variety of machine learning algorithms (such as support vector machine, random forest, deep learning, etc.) are used for modeling. In order to improve the training efficiency and prediction accuracy of the model, the optimization algorithm (such as whale algorithm) can be used to optimize the model parameters. The training set is input into the optimized machine learning model for training until the model reaches the predetermined performance standard. Then, the trained model is verified using the test set to evaluate the model's performance indicators such as prediction accuracy and stability. The data of the area to be evaluated is input into the trained geological disaster risk evaluation model for risk prediction. According to the prediction results, the geological disaster risk level classification standard is formulated, and the prediction results are displayed in the form of a geological disaster risk distribution map. Based on the actual situation in the region, formulate targeted geological disaster prevention and control strategies and emergency plans to provide a scientific basis for decision-making by the government and relevant departments.

[0074] The effects of the above technical solutions are as follows: by acquiring multi-source data such as original remote sensing images, topographic data, geological structure data and meteorological data of the target area, a comprehensive and rich information basis is provided for the geological hazard risk assessment; the original data is processed using geographic information system software such as ArcGIS to achieve data standardization, grid division and extraction of geological hazard related factor data, thereby improving data processing efficiency and accuracy; according to the occurrence mechanism and regional characteristics of geological disasters, a series of scientific and reasonable susceptibility and vulnerability evaluation indicators are selected, such as slope, slope aspect, elevation, rock and soil type, population density, etc., to construct a comprehensive geological hazard risk assessment indicator system; the evaluation indicators are forward-processed and normalized to eliminate the influence of different dimensions and magnitudes on the evaluation results, making the evaluation results more objective and accurate; a variety of machine learning algorithms such as support vector machines, random forests, and deep learning are used for modeling, making full use of the advantages of different algorithms and improving the generalization ability and prediction accuracy of the model; using whale The optimization algorithms such as the fish algorithm optimize the parameters of the support vector machine model, reduce the model training time, improve the model training efficiency and prediction accuracy; input the data of the area to be evaluated into the trained geological hazard risk assessment model to predict the risk, providing scientific and accurate prediction results; formulate the geological hazard risk classification standards based on the prediction results, and draw the geological hazard risk distribution map, which intuitively shows the spatial distribution characteristics of geological hazard risks; formulate targeted geological hazard prevention and control strategies and emergency plans based on the actual regional conditions, and provide strong support for the decision-making of the government and relevant departments; through the geological hazard risk assessment, potential geological hazard risks can be warned in advance, providing a time window for disaster prevention and emergency response; it helps to optimize the allocation of geological hazard prevention and control resources, invest limited resources in higher risk areas, and improve prevention and control effects; based on scientific evaluation methods and prediction results, it helps the government and relevant departments make more scientific and reasonable decisions and promote the effective implementation of geological hazard prevention and control work.

[0075] In summary, this method significantly improves the accuracy and scientificity of geological hazard assessment by integrating multi-source data, constructing a scientific and reasonable evaluation index system, adopting efficient machine learning modeling and optimization technology, and providing accurate prediction and decision-making support, providing strong technical support for geological disaster prevention and control work.

[0076] In one embodiment of the present invention, the S1 includes:

[0077] S11. Determine the specific objectives of geological hazard assessment, including regional scope, assessment accuracy requirements, time scale, etc.; determine the required multi-source data types and accuracy requirements based on historical disaster records, geological exploration reports and expert consultation;

[0078] S12. Collect original remote sensing images through various methods (satellite remote sensing, drone aerial photography, ground measurement), and use GIS software to perform geometric correction, radiation correction and atmospheric correction on remote sensing images;

[0079] S13. Obtain topographic data and geological structure data from the geological department and perform data consistency verification. Obtain long-term and short-term meteorological data from the meteorological department, including rainfall, wind direction and speed, etc.;

[0080] S14. Use data fusion technology to align multi-source data in space and time;

[0081] S15. Based on the evaluation requirements and data resolution, an adaptive grid division method is used to divide the target area into grid cells of varying sizes but interconnected;

[0082] S16. For each grid unit, use GIS spatial analysis tools to extract geological hazard related factor data.

[0083] The working principle of the above technical scheme is: clarify the specific goals of geological hazard assessment, including regional scope, assessment accuracy requirements, time scale, etc. These goals will guide the direction of subsequent data collection and processing; based on historical disaster records, geological exploration reports and expert consultation, determine the required multi-source data types and accuracy requirements. This includes remote sensing images, topographic data, geological structure data, and meteorological data; collect original remote sensing images through various means (such as satellite remote sensing, drone aerial photography, and ground measurement); use GIS software to perform geometric correction, radiation correction, and atmospheric correction on remote sensing images to eliminate image distortion, improve image quality, and ensure the accuracy of subsequent analysis; obtain topographic data and geological structure data from the geological department, and perform data consistency verification to ensure the accuracy and reliability of the data; obtain long-term and short-term meteorological data from the meteorological department, including rainfall, wind direction and wind speed, etc., which are crucial for analyzing the inducing factors of geological disasters; through data fusion technology, multi-source data are spatially and temporally aligned to ensure that data from different sources can be analyzed and compared in the same space-time framework; according to evaluation requirements and data resolution, an adaptive grid division method is used to divide the target area into grid units of varying sizes but interconnected. This method can more flexibly reflect the complexity of topography and geological disasters in the region; for each grid unit, GIS spatial analysis tools are used to extract data on geological disaster-related factors. These factor data include slope, aspect, elevation, rock and soil type, river density, vegetation coverage, etc., which are of great significance for assessing the susceptibility and danger of geological disasters.

[0084] The effects of the above technical solutions are as follows: by clarifying the specific objectives of geological hazard assessment, including regional scope, assessment accuracy requirements, time scale, etc., a clear guiding direction is provided for subsequent data collection and processing; based on historical disaster records, geological exploration reports and expert consultations, the required multi-source data types and accuracy requirements are accurately determined to ensure the pertinence and effectiveness of the data; original remote sensing images are collected by various means such as satellite remote sensing, drone aerial photography, and ground measurement, which improves the efficiency and comprehensiveness of data collection; GIS software is used to perform geometric correction, radiation correction and atmospheric correction on remote sensing images, effectively eliminating image distortion, improving image quality, and providing a reliable data basis for subsequent analysis; through data fusion technology, multi-source data are spatially and temporally aligned, realizing the organic integration of data and improving the integrity and consistency of data; consistency checks are performed on data obtained from geological and meteorological departments to ensure the accuracy and reliability of the data and avoid evaluation biases caused by data errors; according to evaluation requirements and data resolution, an adaptive grid division method is used to divide the target area into grid units of varying sizes but interconnected. This method not only takes into account the complexity of the topography and landforms in the region, but also improves the precision and pertinence of the analysis; for each grid unit, GIS spatial analysis tools are used to extract geological hazard-related factor data, providing detailed data support for subsequent risk assessment. These factor data can fully reflect the potential risks and influencing factors of geological hazards; by integrating multi-source data, a variety of influencing factors of geological hazards are comprehensively considered, which improves the scientificity and comprehensiveness of the evaluation; on the basis of grid division and factor data extraction, each grid unit can be finely analyzed and evaluated, which improves the accuracy and precision of the evaluation.

[0085] In summary, the above-mentioned technical scheme significantly improves the scientificity, accuracy and comprehensiveness of the evaluation in the S1 stage of geological hazard risk assessment through measures such as clear goals and precise data acquisition, efficient data collection and processing, data integration and consistency verification, flexible grid division and detailed analysis, etc., providing strong support for subsequent disaster prevention and mitigation work.

[0086] In one embodiment of the present invention, the S14 includes:

[0087] Standardize the format, coordinate system and projection of data from different sources (such as satellite remote sensing, drone aerial photography, ground measurement, geological exploration, meteorological monitoring, etc.) to ensure that all data are analyzed in the same reference frame; use statistical methods and image processing techniques such as median filtering and Gaussian filtering to remove noise and outliers in the data;

[0088] Using algorithms such as SIFT (Scale Invariant Feature Transform) and SURF (Speeded Up Robust Features), we can extract stable feature points from remote sensing images, topographic maps, and geological maps, and accurately match them to establish spatial correspondence between different data sources.

[0089] Based on the feature point matching results, the affine transformation model is used to adjust the relative position of the data set, and the gaps caused by the transformation are filled by appropriate interpolation methods (such as bilinear interpolation and bicubic interpolation) to maintain the continuity and integrity of the data;

[0090] For time series data such as meteorological data and geological disaster historical records, a complete time series framework is constructed to accurately correspond the data at each time point to the evaluation system; considering the differences in update frequencies of different data sources, time interpolation (such as linear interpolation and Lagrange interpolation) is used to adjust the time series of each data source to a unified time resolution;

[0091] According to the reliability, resolution and timeliness of the data source, different weights are assigned to each data type, and data fusion is performed through weighted averaging.

[0092] The working principle of the above technical solution is: standardize the format, coordinate system and projection of data from different sources (such as satellite remote sensing, drone aerial photography, ground measurement, geological exploration, meteorological monitoring, etc.). This is to ensure that all data can be analyzed in the same reference frame to avoid errors caused by inconsistent formats or coordinate systems; use statistical methods and image processing techniques (such as median filtering, Gaussian filtering, etc.) to remove noise and outliers in the data. These technologies can smooth data, reduce random errors, and improve data quality; use algorithms such as SIFT (Scale Invariant Feature Transform) and SURF (Speeded Robust Features) to extract stable feature points from remote sensing images, topographic maps, and geological maps. These feature points are scale-invariant and rotation-invariant, and can maintain consistency at different scales and rotation angles; accurately match the extracted feature points to establish spatial correspondence between different data sources. This helps to determine the relative position relationship between different data sources and provides a basis for subsequent data fusion; based on the feature point matching results, the affine transformation model is used to adjust the relative position of the data set. Affine transformation can maintain the linear relationship and relative proportion of data, so that different data sources can be aligned in space; fill the gaps caused by transformation through appropriate interpolation methods (such as bilinear interpolation and bicubic interpolation). This helps to maintain the continuity and integrity of data and avoid holes or breaks in the data fusion process; build a complete time series framework for time series data such as meteorological data and geological disaster historical records. This helps to accurately correspond the data at each time point to the evaluation system, which is convenient for subsequent comprehensive analysis; considering the differences in the update frequency of different data sources, time interpolation (such as linear interpolation and Lagrange interpolation) is used to adjust the time series of each data source to a unified time resolution. This helps to eliminate time inconsistencies and enable different data sources to be effectively integrated in the time dimension; assign different weights to each data type based on factors such as data source reliability, resolution and timeliness. Weight allocation reflects the importance and credibility of different data sources in the evaluation process; fuse data through weighted average method. This method can comprehensively consider the advantages and disadvantages of different data sources and obtain more comprehensive and accurate evaluation results.

[0093] In summary, the technical solution of the S14 stage achieved the effective fusion and integration of multi-source data through steps such as data standardization, data cleaning and denoising, feature extraction and matching, data set position adjustment and interpolation, time series data integration, and data fusion and weight allocation, providing high-quality data support for subsequent geological hazard assessment.

[0094] The effects of the above technical solutions are as follows: through standardization processing, it is ensured that data from different sources can be analyzed in the same reference frame, eliminating errors caused by inconsistent formats, coordinate systems and projections, and improving the consistency and comparability of data; using statistical methods and image processing technology to remove noise and outliers in the data, significantly improving data quality and reducing the impact of data errors on evaluation results; using algorithms such as SIFT and SURF to extract stable feature points from remote sensing images, topographic maps and geological maps, and accurately matching them, establishing spatial correspondence between different data sources. This provides a solid foundation for subsequent data fusion and comprehensive analysis; based on the feature point matching results, the affine transformation model is used to adjust the relative position of the data set, and the gaps caused by the transformation are filled by interpolation methods to maintain the continuity and integrity of the data. This helps to avoid data breaks or gaps during data fusion; builds a complete time series framework, accurately corresponds the data at each time point to the evaluation system, and facilitates time series analysis; adjusts the time series of different data sources to a unified time resolution through time interpolation, eliminates time inconsistencies, and enables different data sources to be effectively integrated in the time dimension; assigns different weights to each data type based on factors such as data source reliability, resolution, and timeliness, and performs data fusion through weighted averaging. This method can comprehensively consider the advantages and disadvantages of different data sources and obtain more scientific and accurate evaluation results; after the data fusion processing of the S14 stage, the multi-source data required for geological hazard risk assessment has been effectively integrated and uniformly processed, providing a high-quality data foundation for subsequent comprehensive analysis and decision support. This helps to improve the efficiency and accuracy of the evaluation and provide a more reliable basis for disaster prevention and mitigation work.

[0095] In summary, the technical solution of the S14 stage has played an important role in the geological hazard assessment. Through methods such as data standardization, quality improvement, establishment of spatial correspondence, maintenance of data continuity and integrity, time series data integration and scientific data fusion, it has significantly improved the scientificity, accuracy and efficiency of the assessment.

[0096] In one embodiment of the present invention, the S15 includes:

[0097] Based on the overall size of the assessment area and the data resolution, an initial benchmark grid size is set; this size should roughly reflect the basic characteristics of the assessment area and the preliminary distribution of geological hazard risks; using GIS software, the assessment area is preliminarily evenly divided according to the benchmark grid size to generate a preliminary grid framework;

[0098] Use topographic data (such as DEM data) to analyze the terrain complexity of the evaluation area, including slope, aspect, elevation change, etc.; combine historical disaster records, geological exploration reports and expert knowledge base to identify high-risk and low-risk areas for geological disasters;

[0099] In high-risk areas or areas with complex terrain, the grid size is reduced according to actual needs to achieve more refined division; in low-risk areas or areas with flat terrain, the grid size is increased to improve calculation efficiency;

[0100] In addition to adjusting the grid size, the shape of the grid can also be adjusted according to the terrain characteristics and geological disaster risk distribution to make it better fit the actual terrain and potential disaster boundaries.

[0101] During the meshing process, there may be overlaps or gaps at the mesh edges. Through algorithm optimization, we ensure that the meshes are seamlessly connected and the overlaps are minimized.

[0102] Check whether the data resolution in each grid unit meets the evaluation requirements; for areas with insufficient resolution, further collect high-resolution data or use data interpolation methods to improve data quality;

[0103] Use GIS spatial analysis tools to evaluate the quality of the divided grids, including the regularity of grid shape, size consistency, data coverage integrity, etc., assign a unique code or identifier to each grid unit, and establish a grid unit attribute database to record the key data of each grid; the key data includes basic information (such as location, size, shape), terrain characteristics and geological disaster risks.

[0104] The working principle of the above technical solution is as follows: based on the overall size of the evaluation area and the data resolution, an initial benchmark grid size is set. This size should roughly reflect the basic characteristics of the evaluation area and the preliminary distribution of geological disaster risks; using GIS software, the evaluation area is preliminarily and evenly divided according to the benchmark grid size to generate a preliminary grid framework. This step provides a basis for subsequent fine-tuning; using topographic and geomorphic data (such as DEM data), the terrain complexity of the evaluation area is analyzed, including slope, slope aspect, elevation changes, etc. These terrain features have an important impact on the occurrence of geological disasters; combining historical disaster records, geological exploration reports and expert knowledge bases, high-risk areas and low-risk areas for geological disasters are identified. This helps to give different attention in subsequent grid division; in high-risk areas or areas with high terrain complexity, the grid size is reduced according to actual needs to achieve a more refined division. This helps to capture subtle changes in geological hazards more accurately; in low-risk areas or areas with gentle terrain, the grid size is increased to improve computational efficiency. This differentiated processing makes grid division more reasonable and efficient; in addition to adjusting the grid size, the shape of the grid can also be adjusted according to the terrain characteristics and the distribution of geological hazard risks to better fit the actual terrain and potential disaster boundaries. This helps to more accurately reflect the spatial distribution characteristics of geological hazards; in the process of grid division, grid edge overlap or gaps may occur. Through algorithm optimization, ensure that the grids are seamlessly connected and the overlap is minimized. This helps to maintain the continuity and integrity of the data; check whether the data resolution within each grid cell meets the evaluation requirements. For areas with insufficient resolution, further collect high-resolution data or use data interpolation methods to improve data quality. This helps to ensure the accuracy and reliability of the evaluation results; use GIS spatial analysis tools to evaluate the quality of the divided grid, including grid shape regularity, size consistency, data coverage integrity, etc. This helps to ensure the scientificity and rationality of grid division; assign a unique code or identifier to each grid cell, and establish an attribute database for the grid cell. The database records the key data of each grid, including basic information (such as location, size, shape), terrain characteristics, and geological hazard risks. This helps with subsequent data management and analysis.

[0105] The effects of the above technical solutions are as follows: based on the overall size and data resolution of the evaluation area, a reasonable initial benchmark grid size is set, which can roughly reflect the basic characteristics of the evaluation area and the preliminary distribution of geological disaster risks; in high-risk areas or areas with high terrain complexity, the grid size is reduced to achieve a more refined division, which helps to more accurately capture the slight changes in geological hazards and improve the accuracy of the evaluation; in low-risk areas or areas with flat terrain, the grid size is increased, which reduces unnecessary calculations and improves calculation efficiency. At the same time, this also helps to optimize the use of data resources and avoid wasting too many resources in unimportant areas; the shape of the grid is adjusted according to the terrain characteristics and the distribution of geological disaster risks to better fit the actual terrain and potential disaster boundaries. This flexibility makes the grid division more in line with the actual situation and improves the reliability of the evaluation results; by checking the data resolution in each grid unit, it is ensured that the evaluation requirements are met. For areas with insufficient resolution, further collection of high-resolution data or data interpolation methods are adopted to improve data quality, thereby ensuring the quality of the data based on which the evaluation is based; GIS spatial analysis tools are used to evaluate the quality of the divided grid, including grid shape regularity, size consistency, data coverage integrity and other aspects. This helps ensure the scientificity and rationality of grid division and the accuracy of subsequent evaluation; assign a unique code or identifier to each grid unit and establish a grid unit attribute database. This helps to quickly locate and query relevant data in subsequent data management and analysis work, improving work efficiency. At the same time, recording the key data of each grid (such as basic information, terrain characteristics, geological disaster risks, etc.) also provides a comprehensive and detailed information basis for comprehensive evaluation.

[0106] In summary, this technical solution not only improves the accuracy and efficiency of geological hazard assessment through scientific and reasonable grid division and refined data processing and analysis methods, but also ensures the reliability and practicality of the assessment results. This is of great significance for guiding geological disaster prevention and control work and protecting people's lives and property.

[0107] In one embodiment of the present invention, an initial reference grid size is set based on the overall size of the evaluation area and the data resolution; the evaluation area is preliminarily and evenly divided according to the reference grid size using GIS software to generate a preliminary grid framework; including:

[0108] Use GIS software to accurately measure the overall size of the evaluation area, including length, width and area, to provide a basis for setting the benchmark grid size; analyze the resolution of existing data, including the pixel size or spatial resolution of remote sensing images, topographic data, etc., to determine its impact on grid division;

[0109] Ensure that the data coverage in the evaluation area is uniform, with no significant data gaps or missing areas, laying the foundation for the fairness of grid division.

[0110] Based on the evaluation area size and data resolution, a reasonable benchmark grid size is preliminarily calculated using mathematical formulas or empirical rules; this calculation should take into account the evaluation accuracy requirements and computing resource limitations.

[0111] Based on the expert knowledge base, the size of the preliminary calculated benchmark grid is adjusted to ensure that it meets the accuracy requirements but is not too large;

[0112] In a small number of representative areas, grids of different sizes are used for simulation division to evaluate their impact on the results of geological hazard assessment and select the optimal benchmark grid size.

[0113] Use the grid division tool of GIS software to make a preliminary and uniform division of the evaluation area according to the set benchmark grid size; ensure that the grid boundary is consistent with the regional boundary and there is no overlap or omission between grids.

[0114] Check the preliminary grid boundaries to ensure their rationality in complex terrain or important geological structures (such as river confluences, fault zones, etc.); if necessary, manually adjust the grid boundaries to adapt to actual conditions.

[0115] The working principle of the above technical solution is as follows: using GIS software, accurately measuring the overall size (length, width, area) of the evaluation area through spatial analysis functions, which is the basis for setting the benchmark grid size; analyzing the resolution of existing data, especially the pixel size or spatial resolution of key data such as remote sensing images and topographic data. The resolution of the data directly affects the detail and accuracy of the data in each grid unit after grid division; ensuring uniform data coverage in the evaluation area and avoiding data blanks or missing areas is an important prerequisite for ensuring the fairness of grid division and the accuracy of subsequent evaluation; based on the evaluation area size and data resolution, using mathematical formulas or empirical rules to preliminarily calculate a reasonable benchmark grid size. This calculation process needs to comprehensively consider the accuracy requirements of the evaluation and the limitations of computing resources to balance the accuracy and computing efficiency of the evaluation results; with the help of the expert knowledge base, the preliminarily calculated benchmark grid size is adjusted. The expert knowledge base contains professional knowledge and experience in the fields of geology, GIS and disaster assessment, which can ensure that the adjusted grid size meets the accuracy requirements and is not too large, thereby reducing the amount of calculation while ensuring the evaluation accuracy; by performing simulation division in a small number of representative areas, grids of different sizes are used to evaluate their impact on the results of geological hazard assessment. This process helps to select the optimal benchmark grid size, that is, the grid size that can most accurately reflect the risk of geological hazards; the grid division tool of the GIS software is used to preliminarily divide the evaluation area evenly according to the set benchmark grid size. This step ensures the uniform distribution of the grid in the entire evaluation area, which facilitates subsequent data processing and analysis; the boundaries of the preliminarily divided grids are checked, especially in complex terrain or important geological structures (such as river confluences, fault zones, etc.). The geological conditions in these areas are complex and have an important impact on the occurrence of geological hazards. Therefore, it is necessary to ensure the rationality of the grid boundaries in these areas, and the grid boundaries can be manually adjusted to adapt to the actual situation when necessary; after the above steps, a preliminary grid framework is generated. This framework is the basis for the subsequent geological hazard assessment, which is used to divide the evaluation area into several grid cells with clear boundaries and unique codes; each grid cell contains key data related to the risk of geological hazards (such as slope, slope aspect, rock and soil type, etc.), which will be used for the subsequent evaluation model construction and hazard prediction.

[0116] The effects of the above technical solution are: by accurately measuring the overall size of the evaluation area and analyzing the data resolution, a reasonable benchmark grid size can be set. This ensures that the data in each grid unit is sufficiently representative, thereby improving the accuracy of the geological hazard assessment; the grid boundaries are carefully checked and adjusted at complex terrain or important geological structures to ensure that these key areas are fully considered in the evaluation process, further improving the accuracy of the evaluation; when preliminarily calculating the benchmark grid size, the accuracy requirements of the evaluation and the limitations of computing resources are comprehensively considered. This avoids excessive calculations due to too fine a grid, thereby optimizing the utilization of computing resources; by performing simulated divisions in a small number of representative areas and selecting the optimal benchmark grid size, the efficient utilization of computing resources is further ensured; ensuring that the data coverage in the evaluation area is uniform, without significant data gaps or missing areas, laying the foundation for the fairness of grid division. This avoids evaluation bias caused by uneven data distribution; the uniform division of the grid and the rationality check of the boundaries ensure that each grid unit is treated equally during the evaluation process, further enhancing the fairness of the evaluation. The use of the grid division tool of the GIS software for automated division greatly improves the efficiency of grid division. Compared with manual division, this method is faster and more accurate. By simulating division and evaluation in a small number of representative areas, the optimal benchmark grid size is quickly selected, which reduces unnecessary trial and error and further improves the evaluation efficiency. The generated grid framework provides a solid foundation for subsequent data processing, model building and hazard prediction. Each grid unit contains key data related to the risk of geological hazards, which facilitates in-depth analysis and application. The standardization and normalization of the grid framework makes the geological hazard evaluation results in different regions and at different time points comparable, providing strong support for regional geological hazard prevention and control planning and decision-making.

[0117] In summary, this technical solution has many benefits and effects in geological hazard risk assessment, including improving evaluation accuracy, optimizing computing resource utilization, enhancing evaluation fairness, improving evaluation efficiency, and supporting subsequent analysis and application.

[0118] In one embodiment of the present invention, the GIS spatial analysis tool is used to evaluate the quality of the divided grids, assign a unique code or identifier to each grid unit, and establish an attribute database of the grid unit to record the key data of each grid, including:

[0119] Use the boundary analysis tool of GIS software to check the boundary of each grid unit to see if it is complete and continuous, and to ensure that there is no overlap or omission; establish the topological relationship between grid units, including adjacent relationships, inclusion relationships, etc., to provide a basis for subsequent spatial analysis;

[0120] Check whether each grid cell contains all necessary data on geological hazards related factors, such as slope, aspect, rock and soil type, etc.; check the consistency and rationality of the data in the grid cell through cross-validation and comparison with historical data, and exclude outliers or erroneous data;

[0121] According to the data resolution and grid size, assess whether the data accuracy meets the requirements of geological hazard assessment; for data with insufficient accuracy, use data fusion, interpolation or higher-precision data sources to supplement;

[0122] According to the geological hazard risk assessment index system, the attribute fields of the grid units are designed, including but not limited to geological hazard susceptibility indicators (such as slope, slope aspect, river density, etc.), vulnerability indicators (such as population density) and special geological condition identifiers (such as faults, karst, etc.), and the data of each grid unit is entered into the attribute database and a secondary verification is performed;

[0123] Establish the association relationship between attribute fields, such as conducting association analysis between geological hazard susceptibility index and vulnerability index; at the same time, create an index for the database, and use the mapping function of GIS software to display the grid units and their attribute data in the form of a map, so as to facilitate intuitive understanding of the distribution of geological hazards in the evaluation area.

[0124] The working principle of the above technical solution is: using the boundary analysis tool of GIS software, the boundary of each grid unit is automatically checked to ensure that the boundary is complete, continuous, without overlap and omission. This is the basic step of grid division quality assessment, providing an accurate grid framework for subsequent spatial analysis; establishing topological relationships between grid units, including adjacent relationships, inclusion relationships, etc. These topological relationships are crucial for understanding the spatial connection between grid units and conducting spatial analysis (such as buffer analysis, overlay analysis, etc.); checking whether each grid unit contains all necessary geological hazard related factor data, such as slope, aspect, rock and soil type, etc. This is a key step to ensure the accuracy of the evaluation results, because any missing data may lead to deviations in the evaluation results; checking the consistency and rationality of the data in the grid unit by cross-validation and comparison of historical data. Cross-validation can help discover potential errors or outliers in the data, while comparison of historical data can verify the accuracy and reliability of the data; according to the data resolution and grid size, assess whether the accuracy of the data meets the needs of geological hazard assessment. If the data accuracy is insufficient, data fusion, interpolation or higher-precision data sources will be used to supplement it to improve the accuracy of the evaluation results; according to the geological hazard risk evaluation index system, the attribute fields of the grid unit are designed. These fields include but are not limited to geological hazard susceptibility indicators, vulnerability indicators, and special geological condition identifiers, which are used to comprehensively describe the geological hazard hazard characteristics of the grid unit; the data of each grid unit is entered into the attribute database and a secondary check is performed to ensure the accuracy and integrity of the data. Secondary verification is an important step to ensure data quality, which can further reduce the impact of data errors and outliers; establish associations between attribute fields, such as associating geological hazard susceptibility indicators with vulnerability indicators. This association analysis helps to deeply understand the formation mechanism and influencing factors of geological hazard risks; create indexes for the database to improve the efficiency of data retrieval and query. Indexes can speed up data access, especially when processing large amounts of data, and can significantly improve the performance of the database; use the mapping function of GIS software to display grid units and their attribute data in the form of maps. This visualization method facilitates intuitive understanding of the distribution of geological hazards in the evaluation area and provides an intuitive reference for decision makers.

[0125] In summary, through the application of GIS spatial analysis tools, this technical solution realizes multiple links such as quality assessment of divided grids, inspection of data integrity and rationality, assessment and supplement of data accuracy, establishment of attribute database and data entry, association of attribute fields and database optimization, and map display and visualization, providing comprehensive, accurate and efficient technical support for geological hazard risk assessment.

[0126] The effects of the above technical solutions are as follows: through the boundary analysis tool of GIS software, it can ensure that the boundaries of each grid unit are complete, continuous, without overlap and omission, thereby improving the accuracy and reliability of grid division; establishing topological relationships between grid units (such as adjacent relationships and inclusion relationships) provides a solid foundation for subsequent spatial analysis, which helps to more accurately understand the spatial connections between grid units; ensuring that each grid unit contains all necessary geological hazard-related factor data (such as slope, aspect, rock and soil type, etc.) to prevent the impact of data missing on the evaluation results; through cross-validation and comparison of historical data, outliers or erroneous data in grid units can be excluded to improve data accuracy and reliability; data accuracy is evaluated based on data resolution and grid size, and data fusion, interpolation or higher-precision data sources are used to supplement data with insufficient accuracy, thereby improving the accuracy of geological hazard risk assessment; according to The geological hazard risk assessment index system designs the attribute fields of the grid unit, realizes the normalization and standardization of data management, and facilitates the subsequent data analysis and application; the data of each grid unit is entered into the attribute database, and a secondary verification is performed to ensure the accuracy and completeness of the data and reduce the impact of data errors on the evaluation results; the establishment of the association between the attribute fields (such as the association analysis between the geological hazard susceptibility index and the vulnerability index) helps to deeply understand the formation mechanism and influencing factors of the geological hazard risk and improve the scientificity and accuracy of the analysis; creating an index for the database can significantly improve the efficiency of data retrieval and query, speed up data processing, and meet the timeliness requirements of geological hazard risk assessment; using the mapping function of GIS software, the grid unit and its attribute data are displayed in the form of a map, which is convenient for intuitively understanding the distribution of geological hazards in the evaluation area and providing an intuitive reference for decision makers.

[0127] In summary, this technical solution provides strong technical support and guarantee for geological hazard assessment by improving grid division quality, ensuring data integrity and rationality, improving data accuracy, standardizing data management, improving analysis efficiency and effectiveness, and intuitive display and evaluation.

[0128] In one embodiment of the present invention, the S2 includes:

[0129] S21. Organize a multidisciplinary expert team in geology, remote sensing, meteorology, etc. to conduct an in-depth analysis of the susceptibility of geological disasters; based on literature research and expert experience, preliminarily screen out a series of potential susceptibility evaluation indicators;

[0130] S22. Use correlation analysis and principal component analysis to reduce the dimension and rank the importance of the indicators initially screened to determine the key evaluation indicators;

[0131] S23. Analyze the density of human activities and potential economic losses in the region by combining socio-economic data (e.g., census data, transportation network maps, etc.); consider the differences in vulnerability of different socio-economic groups (e.g., residents, enterprises, infrastructure), and develop differentiated vulnerability assessment indicators;

[0132] S24. Based on the analytic hierarchy process (AHP), a geological hazard evaluation index system is constructed, and the evaluation indicators are scored and weighted based on the expert knowledge base, and the weight setting is optimized through multiple rounds of iterations.

[0133] The working principle of the above technical scheme is: experts in the fields of geology, remote sensing, meteorology, etc. are gathered together to use their professional knowledge and experience in their respective fields to conduct a comprehensive analysis of the susceptibility of geological disasters. This multidisciplinary approach can more comprehensively consider the various factors that cause geological disasters; based on literature research and expert experience, a series of potential evaluation indicators related to the susceptibility of geological disasters are preliminarily screened out. These indicators may include geological structure, topography, rainfall intensity, vegetation cover and other factors; statistical methods (such as correlation analysis) and principal component analysis (PCA) are used to reduce the dimension of the preliminarily screened indicators to remove redundant information while retaining the most important features. This step helps to simplify the problem and improve the efficiency and accuracy of subsequent analysis; through dimensionality reduction, the key evaluation indicators that have the greatest impact on the susceptibility of geological disasters are determined. These indicators will serve as the core components of the subsequent evaluation system; combined with socioeconomic data (such as census data, transportation network maps, etc.), the density of human activities in the evaluation area is quantitatively analyzed. This helps to understand which areas are more vulnerable to geological disasters and assess potential economic losses; considering the differences in vulnerability of different socioeconomic groups (such as residents, enterprises, and infrastructure), differentiated vulnerability evaluation indicators are formulated. These indicators will reflect the vulnerability and resilience of different groups in the face of geological hazards; the analytic hierarchy process (AHP), a scientific evaluation method, is used to construct a geological hazard evaluation index system. AHP decomposes complex problems into several levels and factors, and compares the elements of the same level in pairs to quantitatively describe the degree of importance, thereby calculating the weights that reflect the relative importance order of the elements at each level, and calculating the weights of the relative importance order of all elements through the total ranking between all levels; the evaluation indicators are scored and weighted based on the expert knowledge base. Experts judge the importance of each evaluation indicator based on their own professional knowledge and experience, and give corresponding weight values. The weight setting is optimized through multiple rounds of iteration to ensure the scientificity and rationality of the evaluation system.

[0134] The effects of the above technical solutions are: organizing a multidisciplinary team of experts in geology, remote sensing, meteorology, etc. to conduct in-depth analysis of the susceptibility of geological disasters, which can comprehensively consider multiple factors such as geological structure, topography, climate conditions, etc., so as to more comprehensively evaluate the susceptibility of geological disasters; experts in different fields can each play their expertise, provide targeted insights and analysis, and ensure the accuracy and reliability of the evaluation results; using correlation analysis and principal component analysis to reduce the dimension of the indicators initially screened, remove redundant information, retain the most important features, simplify the problem, and improve the analysis efficiency; through dimensionality reduction and importance ranking, it can be clear which factors have the greatest impact on the susceptibility of geological disasters, providing an important basis for subsequent evaluation work; combining socio-economic data to analyze the intensity of human activities and potential economic losses in the region, so that the evaluation results are closer to the actual situation and have stronger practical significance; considering the differences in vulnerability of different socio-economic groups and formulating differentiated vulnerability evaluation indicators, it is helpful to more accurately evaluate the impact of geological disasters on different groups and formulate differentiated prevention and control measures. The evaluation index system of geological disaster risk assessment is constructed based on the analytic hierarchy process (AHP). By quantitatively describing the degree of importance, the weights reflecting the relative importance order of the elements at each level are calculated, ensuring the scientificity and rationality of the evaluation system. The evaluation indexes are scored and weighted based on the expert knowledge base, making full use of the professional knowledge and experience of experts and improving the accuracy and authority of the evaluation results. The weight setting is optimized through multiple rounds of iterations to further ensure the stability and reliability of the evaluation system, making the evaluation results more accurate and reliable. The technical solution provides a comprehensive, scientific and reasonable evaluation basis for the prevention and response of geological disasters, which helps decision makers to formulate more accurate and effective disaster prevention and mitigation measures and reduce the impact of geological disasters on human society and economy. By organizing a multidisciplinary expert team to conduct collaborative analysis, it promotes exchanges and cooperation between different disciplines and helps promote the in-depth development of the field of geological disaster research. The publication and publicity of the evaluation results can enhance the public's understanding and awareness of geological disasters and improve the disaster prevention and mitigation capabilities of the whole society.

[0135] In one embodiment of the present invention, S3 includes:

[0136] S31. Based on the temporal distribution, spatial distribution and type distribution of disaster points, representative disaster points are selected from the historical geological disaster database as positive samples;

[0137] S32. Use unsupervised learning methods (such as K-means clustering) to generate potential negative sample points outside the buffer zone; combine expert knowledge base and GIS spatial analysis to verify and screen the generated negative sample points;

[0138] S33. Based on the geological hazard evaluation index system, the evaluation index value of each sample point is extracted, and the sample set is expanded using data enhancement technology (such as noise addition, feature transformation, etc.);

[0139] S34. Use stratified sampling or random sampling methods to divide the sample set into a training set and a test set; and further perform data cleaning and preprocessing on the training set and the test set.

[0140] The working principle of the above technical solution is as follows: First, representative disaster points are selected from the historical geological disaster database as positive samples. This step takes into account the temporal distribution (such as seasonality, periodicity), spatial distribution (such as geographical location, terrain characteristics) and type distribution (such as landslides, debris flows, collapses, etc.) of disaster points to ensure the diversity and representativeness of positive samples; through statistical analysis or expert evaluation, the most valuable disaster points for model training are selected from a large amount of historical data as positive samples; unsupervised learning methods (such as K-means clustering) are used to generate potential negative sample points outside the buffer zone. These negative sample points should be far away from known disaster points and represent relatively safe areas; combined with expert knowledge base and GIS spatial analysis, the generated negative sample points are verified and screened. The expert knowledge base provides an in-depth understanding of the conditions for the occurrence of geological disasters, while GIS spatial analysis can evaluate the topographic features, geological structures and other characteristics of negative sample points, thereby ensuring the validity and rationality of negative sample points; based on the geological disaster risk evaluation index system, the evaluation index value of each sample point (including positive and negative samples) is extracted. These indicator values ​​will be used as input data for model training; data enhancement techniques (such as noise addition, feature transformation, etc.) are used to expand the sample set. Data enhancement can increase the diversity and quantity of samples, which helps to improve the generalization ability and robustness of the model; stratified sampling or random sampling methods are used to divide the sample set into training set and test set. Stratified sampling can ensure that the training set and test set maintain similar distribution characteristics in terms of disaster type, time distribution, etc.; while random sampling is suitable for situations where the sample size is large and the distribution is relatively uniform; further data cleaning and preprocessing are performed on the training set and test set, including removing outliers, processing missing data, standardizing or normalizing feature values, etc. These steps can ensure the integrity and consistency of the data and improve the efficiency and accuracy of model training.

[0141] The effects of the above technical solution are as follows: by considering the time, space and type distribution of disaster points, representative disaster points are selected from the historical geological disaster database as positive samples, ensuring that the positive samples can fully reflect the actual situation of geological disasters; unsupervised learning methods are used to generate potential negative sample points, and expert knowledge base and GIS spatial analysis are combined for verification and screening, which improves the effectiveness and rationality of negative samples. This method not only increases the number of negative samples, but also ensures the diversity and representativeness of negative samples; the sample set is expanded through data enhancement techniques such as noise addition and feature transformation, increasing the diversity and number of samples. This helps the model learn more feature information during the training process and improves the generalization ability and robustness of the model; stratified sampling or random sampling methods are used to divide the sample set into training set and test set, ensuring the consistency of the distribution of the training set and the test set, which helps to more accurately evaluate the performance of the model; further data cleaning and preprocessing are performed on the training set and the test set, outliers are removed, missing data are processed, and standardized or normalized. These steps ensure the integrity and consistency of the data, and improve the efficiency and accuracy of model training; the evaluation index values ​​of sample points are extracted based on the geological disaster risk evaluation index system, and these index values ​​comprehensively reflect the characteristics of multiple aspects of geological disasters. Through the training and learning of machine learning models, the risk of geological disasters can be more accurately assessed; high-quality sample sets and reasonable sample division methods enable the trained models to have better performance, and can more accurately predict the probability and degree of risk of geological disasters, providing strong support for disaster prevention and mitigation work; this technical solution integrates knowledge and technology from multiple disciplines such as geology, remote sensing, meteorology, and GIS, and promotes cross-disciplinary integration and collaborative innovation; through data-driven methods, geological disaster risk assessment provides a more scientific and objective basis for disaster prevention and mitigation work, and helps to improve the scientificity and accuracy of decision-making.

[0142] In summary, technical solution S3 has significant benefits and effects in geological hazard risk assessment. It can improve the representativeness and diversity of samples, enhance the generalization ability of the model, ensure the integrity and consistency of data, improve the accuracy of geological hazard risk assessment, and promote the in-depth development of geological hazard research.

[0143] In one embodiment of the present invention, the S4 includes:

[0144] S41. Evaluate the applicability, advantages and disadvantages of different machine learning algorithms (such as SVM, RF, CNN, etc.) in geological hazard assessment, and combine multiple algorithms based on existing ensemble learning schemes (such as Stacking, Bagging, etc.);

[0145] S42. Design a detailed parameter optimization plan for each machine learning algorithm, and use meta-heuristic optimization algorithms (such as Whale Algorithm WOA, Genetic Algorithm GA, etc.) to globally search and optimize model parameters;

[0146] S43, evaluating the model performance under different parameter combinations through cross-validation, and selecting the optimal parameter combination; inputting the training set into the optimized machine learning model for training;

[0147] S44. Monitor the performance indicators of the model such as loss function value and accuracy in real time during the training process; optimize the model according to the training results; such as adjusting the learning rate, adding regularization terms, etc., to prevent overfitting and underfitting;

[0148] S45. Use the test set to verify the trained model and calculate performance indicators such as accuracy, recall, and F1 score; draw ROC curves and PR curves to analyze the classification performance and generalization ability of the model.

[0149] S46. Interpret and evaluate the model prediction results based on expert knowledge and actual geological hazards.

[0150] The working principle of the above technical solution is as follows: First, the applicability and advantages and disadvantages of different machine learning algorithms (such as SVM, RF, CNN, etc.) in geological hazard assessment are evaluated. This includes considering factors such as the learning ability, generalization ability, and computational complexity of the algorithm; combining multiple algorithms based on existing ensemble learning schemes (such as Stacking, Bagging, etc.). Ensemble learning improves the overall prediction accuracy and stability by combining the prediction results of multiple models and reduces the bias and variance that may exist in a single model; for each machine learning algorithm, a detailed parameter optimization scheme is designed. Parameter optimization is a key link in the model training process. Reasonable parameter settings can significantly improve the performance of the model; meta-heuristic optimization algorithms (such as whale algorithm WOA, genetic algorithm GA, etc.) are used to globally search and optimize model parameters. These algorithms can find the optimal solution in a complex parameter space and improve the efficiency and accuracy of parameter optimization; and the model performance under different parameter combinations is evaluated through cross-validation. Cross-validation is an effective method to evaluate the generalization ability of a model. It divides the data set into multiple subsets and trains and tests them in turn as training sets and test sets; selects the optimal parameter combination, and inputs the training set into the optimized machine learning model for training. The optimal parameter combination can make the model achieve a good fitting effect on the training set while maintaining good generalization ability; monitor the performance indicators such as the loss function value and accuracy of the model in real time during the training process. These indicators can reflect the training status and performance changes of the model; tune the model according to the training results. For example, adjust the learning rate to control the convergence speed of the model, add regularization terms to prevent overfitting, etc. Model tuning is an iterative process that requires continuous attempts and adjustments to achieve the best performance; use the test set to verify the trained model. The test set is a data set independent of the training set, which is used to evaluate the performance of the model on unknown data; calculate performance indicators such as accuracy, recall, F1 score, etc. to comprehensively evaluate the classification performance of the model. At the same time, draw ROC curves and PR curves to further analyze the classification performance and generalization ability of the model; combine expert knowledge and actual geological disaster situations to interpret and evaluate the model prediction results. Expert knowledge can provide an in-depth understanding of the mechanism and influencing factors of geological disasters, which is helpful for reasonable interpretation and evaluation of model prediction results; evaluate the accuracy and reliability of the model prediction results to determine whether it can meet the needs of actual applications. If there are deviations or deficiencies in the model prediction results, the model can be further improved and optimized.

[0151] The effect of the above technical solution is: by evaluating the applicability and advantages and disadvantages of different machine learning algorithms (such as SVM, RF, CNN, etc.), and combining multiple algorithms based on integrated learning schemes (such as Stacking, Bagging, etc.), the advantages of each algorithm can be fully utilized and the deficiencies can be complemented, thereby improving the prediction accuracy and stability of the model. This method can effectively reduce the deviation and overfitting problems that may exist in a single algorithm; for each machine learning algorithm, a detailed parameter optimization scheme is designed, and a meta-heuristic optimization algorithm (such as Whale Algorithm WOA, Genetic Algorithm GA, etc.) is used to search and optimize the model parameters globally. This method can automatically find the optimal parameter combination, avoid the blindness and time-consuming manual parameter adjustment, and significantly improve the performance of the model; evaluate the model performance under different parameter combinations through cross-validation, and select the optimal parameter combination for training. Cross-validation is an effective method to evaluate the generalization ability of a model, which can ensure that the performance of the model on different data sets is relatively stable, thereby improving the practicality and reliability of the model; during the training process, the performance indicators such as the loss function value and accuracy of the model are monitored in real time, and the model is tuned according to the training results. By adjusting the learning rate, adding regularization terms and other measures, the occurrence of overfitting and underfitting can be effectively prevented, so that the model has good generalization ability while maintaining a high accuracy rate; the trained model is verified using the test set, and performance indicators such as accuracy, recall rate, and F1 score are calculated. These indicators can fully reflect the classification performance of the model. At the same time, the ROC curve and PR curve are drawn to further analyze the classification performance and generalization ability of the model, providing strong support for the improvement and optimization of the model; the model prediction results are interpreted and evaluated in combination with expert knowledge and actual geological disaster situations. Expert knowledge can provide an in-depth understanding of the mechanism and influencing factors of geological disasters, which helps to reasonably interpret and evaluate the model prediction results. This method can ensure that the model prediction results are more in line with the actual situation and improve the accuracy and reliability of geological disaster risk assessment.

[0152] In summary, technical solution S4 significantly improves the accuracy, stability and reliability of geological hazard assessment through algorithm combination and integrated learning, parameter optimization, cross-validation, real-time monitoring and tuning, comprehensive model verification and evaluation, and the combination of expert knowledge and actual conditions, providing strong support for the prevention and response to geological disasters.

[0153] In one embodiment of the present invention, S5 includes:

[0154] S51, inputting the data of the area to be evaluated into the trained geological hazard risk assessment model for prediction, outputting the geological hazard risk score or level of each grid unit, and performing spatial visualization display;

[0155] S52. Combine GIS spatial analysis functions to identify high-risk areas and potential disaster points; formulate geological disaster risk classification standards based on prediction results, such as low risk, medium risk, high risk, etc.;

[0156] S53. Use GIS software to draw a geological hazard distribution map (risk map) to visually display the spatial distribution characteristics of regional geological hazard risks;

[0157] S54. Develop specific prevention and control strategies for high-risk areas and potential disaster sites, such as strengthening monitoring and early warning, implementing engineering control, etc.; consider the cost-effectiveness and feasibility of different prevention and control strategies, and develop a comprehensive prevention and control plan;

[0158] S55. Develop a geological disaster emergency response plan, clarify key elements such as evacuation routes and rescue force deployment when a disaster occurs, and organize emergency drills and training activities; revise and improve the emergency response plan based on drill results and actual conditions.

[0159] The working principle of the above technical solution is as follows: input the topography, geological structure, rainfall, vegetation cover, human activity impact and other multi-source data of the area to be evaluated into the trained geological hazard risk assessment model; these data are integrated and managed through the GIS system to ensure the accuracy and integrity of the data; the geological hazard risk assessment model is used to perform complex calculations and analyses based on the input data to predict the possibility or risk of geological disasters occurring in each grid unit in the future; the model outputs the geological hazard risk score or grade of each grid unit, which reflects the probability of occurrence of geological disasters or the potential impact degree; combined with the GIS spatial analysis function, the risk score or grade output by the model is further analyzed. A one-step spatial analysis can identify high-risk areas and potential disaster points; through spatial analysis, we can more intuitively understand the distribution patterns and characteristics of geological hazards in the region; use GIS software to visualize the geological hazard score or level in the form of graphics, charts, etc., to generate a geological hazard hazard distribution map (risk map); the risk map can intuitively display the spatial distribution characteristics of regional geological hazard risks and provide an intuitive reference for decision makers; based on the geological hazard risk prediction results, formulate geological hazard risk level classification standards, such as low risk, medium risk, high risk, etc.; level classification helps to adopt different prevention and control strategies for areas with different risks; formulate specific prevention and control strategies for high-risk areas and potential disaster points. These strategies may include strengthening monitoring and early warning, implementing engineering governance, optimizing the layout of human activities, etc.; when formulating prevention and control strategies, it is necessary to comprehensively consider the cost-effectiveness and feasibility of different strategies to ensure the scientific nature and operability of the prevention and control plans; formulate geological disaster emergency response plans to clarify key elements such as evacuation routes and rescue force deployment when disasters occur; emergency response plans need to be targeted and operational to ensure that they can be responded to quickly and effectively when disasters occur; organize emergency drills and training activities to improve the emergency response capabilities and coordinated combat capabilities of relevant personnel; through drills and training, the feasibility and effectiveness of emergency response plans can be tested, and they can be revised and improved according to actual conditions.

[0160] The effect of the above technical solution is: by inputting the data of the area to be evaluated into the trained geological hazard assessment model, the geological hazard score or level of each grid unit can be output. This method based on big data and machine learning significantly improves the precision and accuracy of prediction; the prediction results are not limited to a single indicator, but take into account a variety of geological, meteorological, human activities and other factors, and realize a comprehensive assessment of the geological hazard; the prediction results are spatially visualized, and the spatial distribution characteristics of regional geological hazard risks are intuitively presented through the geological hazard distribution map (risk map). This display method helps decision makers quickly understand the distribution of disaster risks and provide strong support for decision-making; combined with GIS spatial analysis functions, it can accurately identify high-risk areas and potential disaster points. This provides clear goals and directions for subsequent prevention and control work; specific prevention and control strategies are formulated for high-risk areas and potential disaster points. These strategies may include strengthening monitoring and early warning, implementing engineering governance, and optimizing the layout of human activities. By comprehensively considering the cost-effectiveness and feasibility of different prevention and control strategies, a more scientific and reasonable comprehensive prevention and control plan can be formulated; targeted prevention and control strategies can effectively reduce the probability and impact of geological disasters and improve the efficiency and effectiveness of prevention and control work; formulate geological disaster emergency response plans and clarify key elements such as evacuation routes and rescue force configuration when disasters occur. This will help to quickly activate the emergency response mechanism when a disaster occurs and reduce disaster losses; organize emergency drills and training activities to improve the emergency response capabilities and collaborative combat capabilities of relevant personnel.

[0161] In summary, the above-mentioned technical solution S5 plays an important role in the risk assessment and management of geological disasters. Through scientific prediction, accurate assessment, spatial visualization, formulation of targeted prevention and control strategies, and improvement of emergency response capabilities, it effectively reduces the risks and losses of geological disasters, and provides strong guarantees for regional economic and social development and the safety of people’s lives and property.

[0162] One embodiment of the present invention, as Figure 2 As shown, a geological disaster risk assessment system based on machine learning, the system includes:

[0163] Data acquisition module: determine the target area of ​​geological disasters to be evaluated, obtain multi-source data of the target area (original remote sensing images, topographic data, geological structure data, meteorological data), process the original data based on ArcGIS, divide the grid units, and extract the geological disaster-related factor data of each grid unit;

[0164] System construction module: According to the occurrence mechanism of geological disasters and regional characteristics, geological disaster susceptibility evaluation indicators (slope, slope direction, slope height, river density, elevation, rock and soil type, distance from faults, distance from rivers, vegetation coverage, rainfall) are selected; and geological disaster vulnerability evaluation indicators (population density and economic losses) are determined; the selected evaluation indicators are positively and normalized to construct a geological disaster risk evaluation indicator system;

[0165] Data processing module: select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain the training set and test set;

[0166] Performance evaluation module: Modeling is performed through a variety of machine learning algorithms (such as support vector machines, random forests, deep learning, etc.), and optimization algorithms such as the whale algorithm are used to optimize the parameters of the support vector machine model. The training set is input into the optimized machine learning model for training to obtain a trained geological hazard assessment model; the trained model is verified using the test set to evaluate the model performance;

[0167] Plan formulation module: input the data of the area to be evaluated into the trained geological hazard risk assessment model to make a risk prediction; based on the prediction results, formulate geological hazard risk grade classification standards and draw a geological hazard risk distribution map; based on the actual situation in the region, formulate targeted geological hazard prevention and control strategies and emergency plans.

[0168] The working principle of the above technical solution is: clarify the target area of ​​geological disasters to be evaluated, and widely collect various types of data in the area, including original remote sensing images, topographic data, geological structure data, and meteorological data. These data are the basis for subsequent analysis. Through ArcGIS and other geographic information system software, these original data are processed, such as divided into grid units, to facilitate subsequent data analysis and feature extraction. In each grid unit, extract factor data related to geological disasters, such as slope, slope aspect, elevation, etc.; based on the occurrence mechanism and regional characteristics of geological disasters, select a series of evaluation indicators, which include both geological disaster susceptibility evaluation indicators (such as slope, slope aspect, river density, rock and soil type, etc.) and vulnerability evaluation indicators (such as population density, economic losses, etc.). The selected evaluation indicators are forward-oriented and normalized to ensure that indicators of different dimensions and magnitudes can be compared and calculated in the same evaluation system, so as to construct a complete geological disaster risk evaluation indicator system; in order to train the machine learning model, a sample set needs to be prepared. Historical geological disaster points are usually selected as positive samples, and buffer zones are set around these points to ensure that the model can learn the characteristics of these disaster points. Negative samples are randomly selected outside the buffer zone. The selection of negative samples should ensure the rationality of the spatial range to simulate the situation in non-disaster areas. Based on the geological disaster risk evaluation index system, the sample set is preprocessed, including feature extraction, data cleaning, etc., and finally the training set and test set are obtained. A variety of machine learning algorithms (such as support vector machine, random forest, deep learning, etc.) are used for modeling. In order to improve the training efficiency and prediction accuracy of the model, the optimization algorithm (such as whale algorithm) can be used to optimize the model parameters. The training set is input into the optimized machine learning model for training until the model reaches the predetermined performance standard. Then, the trained model is verified using the test set to evaluate the model's performance indicators such as prediction accuracy and stability. The data of the area to be evaluated is input into the trained geological disaster risk evaluation model for risk prediction. According to the prediction results, the geological disaster risk level classification standard is formulated, and the prediction results are displayed in the form of a geological disaster risk distribution map.

[0169] The effects of the above technical solutions are as follows: by acquiring multi-source data such as original remote sensing images, topographic data, geological structure data and meteorological data of the target area, a comprehensive and rich information basis is provided for the geological hazard risk assessment; the original data is processed using geographic information system software such as ArcGIS, which realizes data standardization, grid division and extraction of geological hazard related factor data, thereby improving data processing efficiency and accuracy; according to the occurrence mechanism and regional characteristics of geological disasters, a series of scientific and reasonable susceptibility and vulnerability evaluation indicators are selected, such as slope, slope aspect, elevation, rock and soil type, population density, etc., to construct a comprehensive geological hazard risk assessment indicator system; the evaluation indicators are positively and normalized to eliminate The influence of different dimensions and magnitudes on the evaluation results makes the evaluation results more objective and accurate; multiple machine learning algorithms such as support vector machines, random forests, and deep learning are used for modeling, which fully utilizes the advantages of different algorithms and improves the generalization ability and prediction accuracy of the model; the support vector machine model is optimized using optimization algorithms such as the whale algorithm, which reduces the model training time and improves the model training efficiency and prediction accuracy; the data of the area to be evaluated is input into the trained geological hazard risk assessment model to perform risk prediction, providing scientific and accurate prediction results; based on the prediction results, a standard for the classification of geological hazard risk levels is formulated, and a geological hazard risk distribution map is drawn, which intuitively shows the spatial distribution characteristics of geological hazard risks.

[0170] In summary, this method significantly improves the accuracy and scientificity of geological hazard assessment by integrating multi-source data, constructing a scientific and reasonable evaluation index system, adopting efficient machine learning modeling and optimization technology, and providing accurate prediction and decision-making support, providing strong technical support for geological disaster prevention and control work.

[0171] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A geological hazard risk assessment method based on machine learning, characterized in that: The method comprises: S1. Determine the target area of ​​geological hazards to be evaluated, obtain multi-source data of the target area, process the original data based on ArcGIS, divide the grid units, and extract the geological hazard related factor data of each grid unit; S2. Select geological disaster susceptibility evaluation indicators based on the occurrence mechanism and regional characteristics of geological disasters; determine geological disaster vulnerability evaluation indicators; perform positive and normalization processing on the selected evaluation indicators to construct a geological disaster risk evaluation indicator system; S3, select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain training sets and test sets; S4. Modeling is performed through a variety of machine learning algorithms. The whale algorithm is used to optimize the parameters of the support vector machine model. The training set is input into the optimized machine learning model for training. The trained model is verified using the test set to evaluate the model performance. S5. Input the data of the area to be evaluated into the trained geological hazard risk assessment model to make risk predictions; formulate geological hazard risk classification standards and draw geological hazard risk distribution maps based on the prediction results; formulate geological hazard prevention and control strategies and emergency plans based on the actual regional conditions; Said S1 comprises: S11. Determine the specific objectives of geological hazard assessment, and determine the required multi-source data types and accuracy requirements based on historical hazard records, geological exploration reports and expert consultation; S12. Collect original remote sensing images through various methods, and use GIS software to perform geometric correction, radiation correction and atmospheric correction on remote sensing images; S13. Obtain topographic data and geological structure data from the geological department and perform data consistency verification, and obtain long-term and short-term meteorological data from the meteorological department; S14. Use data fusion technology to align multi-source data in space and time; S15. Based on the evaluation requirements and data resolution, an adaptive grid division method is used to divide the target area into grid cells of varying sizes but interconnected; S16. For each grid unit, use GIS spatial analysis tools to extract geological hazard related factor data; The S15 comprises: Based on the overall size of the evaluation area and the data resolution, an initial benchmark grid size is set; using GIS software, the evaluation area is preliminarily evenly divided according to the benchmark grid size to generate a preliminary grid framework; Analyze the terrain complexity of the evaluation area using topographic and geomorphic data, and identify high-risk and low-risk areas for geological hazards by combining historical disaster records, geological exploration reports and expert knowledge base; In high-risk areas or areas with complex terrain, the grid size is reduced according to actual needs to achieve a more refined division; in low-risk areas or areas with flat terrain, the grid size is increased; Check whether the data resolution in each grid unit meets the evaluation requirements; for areas with insufficient resolution, further collect high-resolution data or use data interpolation methods to improve data quality; Use GIS spatial analysis tools to evaluate the quality of the divided grids, assign a unique code or identifier to each grid unit, and establish a grid unit attribute database to record the key data of each grid; The GIS spatial analysis tool is used to evaluate the quality of the divided grids, assign a unique code or identifier to each grid unit, and establish an attribute database of the grid unit to record the key data of each grid, including: Using the boundary analysis tool of GIS software, the boundaries of each grid cell are checked and the topological relationship between grid cells is established; Check whether each grid unit contains all the data of geological hazards related factors. Check the consistency and rationality of the data in the grid unit through cross-validation and comparison with historical data, and exclude outliers or erroneous data. According to the data resolution and grid size, assess whether the data accuracy meets the requirements of geological hazard assessment; for data with insufficient accuracy, supplement it through data sources; According to the geological hazard risk assessment index system, the attribute fields of the grid units are designed, the data of each grid unit is entered into the attribute database, and a secondary verification is performed; Establish the association between attribute fields, create an index for the database, and use the mapping function of GIS software to display the grid units and their attribute data in the form of a map.

2. A method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The S14 comprises: Standardize the format, coordinate system and projection of data from different sources, and use statistical methods and image processing techniques to remove noise and outliers in the data; Using SIFT and SURF algorithms, stable feature points are extracted from remote sensing images, topographic maps, and geological maps, and accurate matching is performed to establish spatial correspondence between different data sources; Based on the feature point matching results, the affine transformation model is used to adjust the relative position of the data set, and the gaps caused by the transformation are filled by interpolation method; For time series data, a complete time series framework is constructed to correspond the data at each time point to the evaluation system; time interpolation is used to adjust the time series of each data source to a unified time resolution; According to the reliability, resolution and timeliness of the data source, different weights are assigned to each data type, and data fusion is performed through weighted averaging.

3. A method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The key data include basic information, terrain characteristics and geological disaster risks.

4. The method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The S2 comprises: S21. Conduct in-depth analysis of geological disaster susceptibility; based on literature research and expert experience, preliminarily screen out a series of potential susceptibility evaluation indicators; S22. Use correlation analysis and principal component analysis to reduce the dimension and rank the importance of the indicators initially screened to determine the key evaluation indicators; S23. Analyze the density of human activities and potential economic losses in the region in combination with socio-economic data; formulate differentiated vulnerability assessment indicators; S24. Based on the analytic hierarchy process, a geological hazard risk assessment index system is constructed. The evaluation indicators are scored and weighted based on the expert knowledge base, and the weight settings are optimized through multiple rounds of iterations.

5. The method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The S3 includes: S31. Based on the temporal distribution, spatial distribution and type distribution of disaster points, representative disaster points are selected from the historical geological disaster database as positive samples; S32, using unsupervised learning methods to generate potential negative sample points outside the buffer zone; combining expert knowledge base and GIS spatial analysis to verify and screen the generated negative sample points; S33, based on the geological hazard evaluation index system, extract the evaluation index value of each sample point, and use data enhancement technology to expand the sample set; S34. Use stratified sampling or random sampling methods to divide the sample set into a training set and a test set; and further perform data cleaning and preprocessing on the training set and the test set.

6. A method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The S4 comprises: S41. Evaluate the applicability, advantages and disadvantages of different machine learning algorithms in geological hazard assessment, and combine multiple algorithms based on existing ensemble learning schemes; S42. Design parameter optimization schemes for each machine learning algorithm, and use meta-heuristic optimization algorithms to globally search and optimize model parameters; S43, evaluating the model performance under different parameter combinations through cross-validation, and selecting the optimal parameter combination; inputting the training set into the optimized machine learning model for training; S44. Monitor the performance indicators of the model in real time during the training process; and optimize the model according to the training results; S45. Use the test set to verify the trained model, draw the ROC curve and PR curve, and analyze the classification performance and generalization ability of the model; S46. Interpret and evaluate the model prediction results based on expert knowledge and actual geological hazards.

7. A method for assessing geological hazards based on machine learning according to claim 1, characterized in that: The S5 comprises: S51, inputting the data of the area to be evaluated into the trained geological hazard risk assessment model for prediction, outputting the geological hazard risk score or level of each grid unit, and performing spatial visualization display; S52. Combine GIS spatial analysis functions to identify high-risk areas and potential disaster points; formulate geological disaster risk level classification standards based on prediction results; S53. Use GIS software to draw a geological hazard risk distribution map to intuitively display the spatial distribution characteristics of regional geological hazard risks; S54. Develop specific prevention and control strategies for high-risk areas and potential disaster sites, consider the cost-effectiveness and feasibility of different prevention and control strategies, and develop a comprehensive prevention and control plan; S55. Develop emergency response plans for geological disasters, organize emergency drills and training activities, and revise and improve the emergency response plans based on drill results and actual conditions.

8. A system for implementing the geological hazard risk assessment method based on machine learning as claimed in claim 1, characterized in that: The system comprises: Data acquisition module: determine the target area of ​​geological hazards to be evaluated, obtain multi-source data of the target area, process the original data based on ArcGIS, divide the grid units, and extract the geological hazard related factor data of each grid unit; System construction module: According to the occurrence mechanism and regional characteristics of geological disasters, select geological disaster susceptibility evaluation indicators; determine geological disaster vulnerability evaluation indicators; perform positive and normalization processing on the selected evaluation indicators to construct a geological disaster risk evaluation indicator system; Data processing module: select historical geological disaster points as positive samples, determine the buffer range of positive samples, randomly select negative samples outside the buffer zone, pre-process the sample set based on the geological disaster risk assessment index system, and obtain the training set and test set; Performance evaluation module: Modeling is performed through a variety of machine learning algorithms, the whale algorithm is used to optimize the parameters of the support vector machine model, the training set is input into the optimized machine learning model for training, the trained model is verified using the test set, and the model performance is evaluated; Plan formulation module: input the data of the area to be evaluated into the trained geological hazard risk assessment model to make a risk prediction; based on the prediction results, formulate geological hazard risk grade classification standards and draw a geological hazard risk distribution map; based on the actual situation in the region, formulate geological hazard prevention and control strategies and emergency plans.

Images