A geological disaster identification, monitoring and early warning system and method based on AI technology

Abstract

This invention discloses a geological disaster identification, monitoring, and early warning system and method based on AI technology, relating to the field of geological disaster monitoring technology. The method includes: collecting multi-source datasets from a target area, preprocessing the multi-source datasets, and marking potential hazards to obtain several marker points. Its key technical points are: employing a comprehensive similarity calculation and KNN algorithm to achieve intelligent benchmarking between current high-risk areas and historical disaster cases, achieving intelligent level mapping and improving the scientific rigor and reliability of disaster level determination. By comparing the similarity between current feature values ​​and historical disaster samples, the closest historical case can be quickly located, and the risk level of the current area can be determined accordingly, further enhancing the adaptive capability and response speed of the early warning system. Simultaneously, this method effectively balances the accuracy and response speed of risk assessment.

Description

Technical Field

[0001] This invention relates to the field of geological disaster monitoring technology, specifically to a geological disaster identification, monitoring and early warning system and method based on AI technology. Background Technology

[0002] In recent years, geological disaster monitoring technology has made significant progress with the help of AI technology, especially in the identification and early warning of geological disasters such as landslides and debris flows. Traditional methods mainly rely on field surveys and single sensor data (such as seismographs and rain gauges). This method has problems such as limited coverage, poor real-time performance and insufficient accuracy. AI-based geological disaster identification, through data fusion and machine learning algorithms, has achieved high-precision prediction or early warning of geological disasters.

[0003] Traditional geological hazard monitoring technologies primarily rely on field surveys, single-sensor data (such as seismographs and rain gauges), and static models, which have significant technical limitations. First, the data sources are limited; relying solely on a limited number of ground-based observation devices makes it difficult to comprehensively cover large areas, leading to inaccurate risk assessments. For example, in landslide monitoring, traditional methods may overlook subtle but persistent surface changes that are often crucial for early warning. Second, real-time performance and dynamic adjustment capabilities are insufficient; traditional fixed-parameter models cannot adapt to environmental changes in a timely manner, such as the impact of meteorological conditions on the probability of landslides. This makes the early warning system slow to respond to sudden extreme weather events, reducing the effectiveness of warnings. Furthermore, the methods for setting thresholds lack scientific basis, failing to effectively balance the accuracy of risk assessment with response speed, and also lacking assessment of the chain reaction of disasters. For example, if a landslide of a higher warning level occurs in any adjacent area, whether the resulting vibrations or other factors will affect adjacent areas, and whether the degree of impact will change the original risk level, resulting in inaccurate warning levels, are all technical problems that need to be solved. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A geological disaster identification, monitoring, and early warning method based on AI technology, comprising the following steps:

[0006] Collect multi-source datasets in the target area, preprocess the multi-source datasets, and label potential hazards to obtain several marked points; the multi-source datasets include at least: SAR data, UAV aerial imagery, geographic information, and meteorological data;

[0007] Within the target area, areas with a slope greater than 45° are identified as high-risk areas based on geographic information; based on multi-source datasets and hazard annotations, a feature set is selected for any high-risk area, and a primary risk index is calculated for each high-risk area based on the feature set.

[0008] The GBDT model is used, with meteorological data as input and landslide occurrence probability as output.

[0009] A risk index correction model is constructed to correct the primary risk index based on the probability of landslide occurrence, thereby obtaining the secondary risk index for the corresponding high-risk area.

[0010] The multi-source dataset corresponding to any secondary risk index is compared with similar data from the preceding periods of different historical geological disasters to calculate the comprehensive similarity. A benchmark threshold is set, and all geological disaster event samples in the historical database are traversed to select a number of similar samples. The KNN algorithm is used for level mapping to output the preliminary disaster risk level.

[0011] Analyze the interval between two adjacent sub-risk indices that exceed the standard, determine whether the interval and the two adjacent sub-risk indices that exceed the standard meet the set constraints, and trigger the linkage risk assessment mechanism when they are met to generate a joint risk index; determine whether the joint risk index exceeds the standard index threshold.

[0012] If not, the preliminary disaster risk level will be used as the final disaster risk level;

[0013] If so, then execute the disaster level correction action and send the corresponding warning level based on the final disaster risk level;

[0014] Under the condition that the secondary risk index R_sy continues to accumulate, the probability distribution interval of the secondary risk index R_sy is generated using an AI model; the database used in the KNN algorithm is updated regularly, and the AI ​​model is retrained using new samples; the AI ​​model is any one of GPR, BNN, and random forest.

[0015] Furthermore, the preprocessing of multi-source datasets includes at least: denoising, registration, and spatiotemporal alignment with meteorological data;

[0016] SAR data should include at least: cumulative deformation and deformation rate;

[0017] Drone aerial imagery includes at least: several images of the ground surface;

[0018] Geographic information includes at least: topographic data and slope layers;

[0019] Meteorological data should include at least: rainfall and temperature.

[0020] Furthermore, the feature set includes at least: cumulative deformation, deformation rate, and marker density;

[0021] The primary risk index R_py for each high-risk area is calculated using the following method:

[0022] ;

[0023] In the formula, x i f: The i-th eigenvalue i (): Standardization function, w i : The weights of each feature, i=1,2,...,n, where n is the total number of features.

[0024] Furthermore, the secondary risk index R_sy corresponding to the high-risk area is generated using the following method:

[0025] ;

[0026] In the formula, α: correction coefficient, P_landslide: probability of landslide occurrence;

[0027] Collect the number N_r of all identified high-risk areas within the target area and their corresponding primary risk indices R_py. j Where j = 1, 2, ..., N_r, the primary risk index R_py is calculated. j The mean R_py_avg and standard deviation σ_py;

[0028] The adjustment method for the correction factor α is as follows:

[0029] ;

[0030] In the formula, α0: initial weight; q1 and q2: adjustment coefficients, both ranging from [0, 1]; N_total: number of grids in the target region.

[0031] Furthermore, the comprehensive similarity S_siy is calculated, and the conditions for selecting several similar samples are as follows:

[0032] The overall similarity S_siy is greater than or equal to the benchmark threshold S_th.

[0033] Furthermore, the constraints are as follows:

[0034] Condition 1: Are the geological conditions similar or identical?

[0035] Condition 2: Whether the maximum distance between the high-risk areas of the two sub-risk indices that exceed the standard is lower than the standard value.

[0036] Furthermore, the process of triggering the joint risk assessment mechanism is as follows:

[0037] Construct a joint risk index model, inputting two sub-risk indices that exceed the standard and the geographical distance between the center points of the high-risk areas corresponding to the two sub-risk indices, and outputting the joint risk index R_cd;

[0038] The following methods are used when running the linked risk index model:

[0039] ;

[0040] In the formula, R_sy A and R_sy B : The secondary risk index exceeding the standard for two adjacent high-risk areas; β: the linkage enhancement coefficient, with a value range of [0, 1]; γ: the distance attenuation coefficient, with a value range of [0, 0.1]; d: the geographical distance value.

[0041] Furthermore, the process of implementing disaster level correction is as follows: two adjacent high-risk areas are marked as a whole area, and the initial disaster risk level that is lower is raised by one level to obtain the final disaster risk level.

[0042] A geological disaster identification, monitoring, and early warning system based on AI technology, the system comprising:

[0043] Data acquisition module: Collects multi-source datasets under the target area, preprocesses the multi-source datasets, and marks potential hazards to obtain several marker points; the multi-source datasets include at least: SAR data, UAV aerial imagery, geographic information, and meteorological data;

[0044] Risk identification module: Under the target area, based on geographic information, areas with a slope greater than 45° are identified as high-risk areas; based on multi-source datasets and hazard annotations, a feature set is selected under any high-risk area, and a primary risk index corresponding to each high-risk area is calculated based on the feature set;

[0045] Probability prediction module: Uses GBDT model, takes meteorological data as input and outputs the probability of landslide occurrence;

[0046] Risk Correction Module: Constructs a risk index correction model, corrects the primary risk index based on the probability of landslide occurrence, and derives the secondary risk index for the corresponding high-risk area;

[0047] Comprehensive benchmarking module: It compares the multi-source dataset corresponding to any secondary risk index with similar data from the previous period of different geological disasters in history, calculates the comprehensive similarity; sets a benchmarking threshold, iterates through all geological disaster event samples in the historical database, selects a number of similar samples, uses the KNN algorithm to perform level mapping, and outputs the preliminary disaster risk level;

[0048] Linkage Analysis Module: Analyzes the interval between two adjacent sub-risk indices that exceed the standard, determines whether the interval and the two adjacent sub-risk indices that exceed the standard meet the set constraints, and triggers the linkage risk assessment mechanism when they are met to generate a joint risk index; determines whether the joint risk index exceeds the standard index threshold.

[0049] If not, the preliminary disaster risk level will be used as the final disaster risk level;

[0050] If so, then execute the disaster level correction action and send the corresponding warning level based on the final disaster risk level;

[0051] Adaptive optimization module: Under the condition of continuous accumulation of the secondary risk index R_sy, the AI ​​model is used to generate the probability distribution interval of the secondary risk index R_sy; the database used in the KNN algorithm is updated regularly, and the AI ​​model is retrained with new samples; the AI ​​model is any one of GPR, BNN and random forest.

[0052] This invention provides a geological disaster identification, monitoring, and early warning system and method based on AI technology, which has the following beneficial effects:

[0053] (1) This scheme achieves a comprehensive and effective risk assessment of geological hazard points by using multi-source datasets for fusion, and achieves the effect of high-precision risk identification. This not only improves the detection capability of potential landslide areas, but also solves the problem of inaccurate risk assessment caused by a single data source in traditional methods.

[0054] (2) This scheme uses the GBDT model to predict the probability of landslides and correct the primary risk index. It realizes the dynamic adjustment from the primary risk index to the secondary risk index, and completes the further accurate risk quantification effect. This linkage mechanism can not only take into account geological conditions, but also incorporate the impact of real-time meteorological changes. It effectively solves the problem that traditional fixed parameter models cannot adapt to complex environmental changes, and ensures the timeliness and accuracy of early warning actions.

[0055] (3) The primary risk index generated in this scheme can provide the basic input for risk modeling of subsequent models, enabling them to identify in depth whether there is high geological instability in high-risk areas when considering the influence of meteorological or linkage effects. On the other hand, it can also serve as an important basis for dynamic parameter adjustment, guiding the adaptive adjustment of the correction coefficient α. By statistically analyzing the number of high-risk areas and the distribution of their corresponding primary risk indices, the degree of risk concentration and differences in the entire target area can be determined, thereby determining the weight of the landslide occurrence probability in the calculation of the secondary risk index. This mechanism realizes the transformation from "static assessment" to "dynamic response", improving the sensitivity of the secondary risk index to local high-risk events and the robustness of the overall system.

[0056] (4) This scheme adopts a technical solution of comprehensive similarity calculation and KNN algorithm, which realizes intelligent benchmarking of current high-risk areas and historical disaster cases, achieves the effect of intelligent level mapping, improves the scientificity and reliability of disaster level determination, and can quickly locate the closest historical case by comparing the similarity between the current feature value and the historical disaster sample, and determine the risk level of the current area accordingly, further enhancing the adaptive capability and response speed of the early warning system. Unlike the traditional manual setting of thresholds that emphasizes response speed, and the existing schemes that calculate the required threshold based on historical data to emphasize accuracy, this method can effectively balance the accuracy and response speed of risk determination and meet the actual needs.

[0057] (5) This scheme adopts the joint risk index model and related technologies for modifying disaster levels to achieve quantitative assessment of potential chain reactions between adjacent high-risk areas, thus achieving the effect of regional risk prevention and control closed loop. It solves the shortcomings of traditional isolated analysis of high-risk points that ignores the surrounding impact, making the early warning scheme more comprehensive and flexible. By effectively predicting and responding to possible chain disaster events, the original risk level can be changed, and corresponding early warnings can be sent to attract higher attention, thereby improving the robustness and predictability of the overall scheme. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating the overall steps of a geological disaster identification, monitoring, and early warning method based on AI technology according to the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0060] Example 1:

[0061] Please see Figure 1 This embodiment provides a geological disaster identification, monitoring, and early warning method based on AI technology. This method provides a complete description of the local geological disaster identification and early warning process, covering key steps from data collection, risk modeling, landslide probability prediction, historical similarity analysis to regional linkage effect assessment and AI adaptive optimization. This method is applicable to high-precision geological disaster risk early warning scenarios under the fusion of multi-source remote sensing, meteorological, and geological data, and can provide strong technical support for disaster prevention and mitigation under different conditions.

[0062] The specific steps of this method are explained below:

[0063] S1. Data Acquisition and Preprocessing:

[0064] S1.1 Data Collection:

[0065] Collect multi-source datasets for the target area; the multi-source datasets include at least: SAR data, UAV aerial imagery, geographic information, and meteorological data; form a grid or cell structure for the target area to facilitate subsequent monitoring;

[0066] Remote sensing (SAR) data includes: cumulative deformation and deformation rate in different zones (which may be high-risk areas);

[0067] The drone aerial images include: several surface images;

[0068] Geographic information includes: topographic data (DEM) and slope layers;

[0069] Meteorological data includes: rainfall and temperature in the target area;

[0070] It should be noted that SBAS-InSAR technology is used to analyze surface deformation changes and extract the deformation rate and cumulative deformation under a preset period; for example, the deformation rate is 40 mm / year and the cumulative deformation is 300 mm. High-resolution surface images are collected using drones equipped with camera probes; topographic data is collected online via IoT technology (or LiDAR); GIS software (such as ArcGIS or QGIS) is used to perform topographic analysis on the target area, generating factor layers such as slope and aspect, including the required slope layer; meteorological data such as rainfall and temperature are obtained from meteorological stations on public networks; for multi-source datasets, most of the data can be obtained through on-site data collection by robots, thereby further ensuring the real-time nature and validity of the data.

[0071] S1.2 Hazard Marking:

[0072] Known potential geological hazards in the target area are marked, resulting in several marker points;

[0073] Since this embodiment addresses landslides as the type of geological hazard, it is necessary to mark the landslide locations; for example: mark 18 potential geological hazard points within target area A (usually determined autonomously based on known information);

[0074] S1.3 Data Preprocessing:

[0075] Preprocessing operations are performed on the original multi-source datasets. These preprocessing operations include at least: denoising, registration, and spatiotemporal alignment with meteorological data. For example, the GAMMA software platform is used to perform spatiotemporal baseline connection on SAR data and to synchronize meteorological data with remote sensing images in time.

[0076] S2. High-risk area identification and primary risk index calculation:

[0077] S2.1 Slope Analysis and Identification:

[0078] Within the target area, based on the slope layer in the geographic information, areas with a slope greater than 45° are identified as high-risk areas; the number of such high-risk areas can be one or more.

[0079] Example: Generate a slope layer in ArcGIS, identify high-risk areas with a slope greater than 45°, and select a bounding box in the image to obtain a complete image of an irregular or regular high-risk area;

[0080] S2.2 Summary of Feature Filtering:

[0081] Based on multi-source datasets and hazard annotations, a feature set is selected for any high-risk area; the feature set includes at least: cumulative deformation (in mm), deformation rate (in mm / year), and marker density (in units / km). 2 );

[0082] S2.3, Primary Risk Index Modeling:

[0083] Based on the feature set, the primary risk index R_py corresponding to each high-risk area is calculated using the following method:

[0084] ;

[0085] In the formula, x i The i-th feature value, such as any value of any type in the feature set;

[0086] f i(): Normalization or standardization function used to eliminate dimensional differences; specifically, Min-Max standardization or Z-score normalization can be used. i The weights of each feature value are determined by the AHP (Analytic Hierarchy Process) or the coefficient of variation method; i = 1, 2, ..., n, where n is the total number of feature values ​​participating in the modeling. In this embodiment, the feature set is: cumulative deformation, deformation rate, and marker density, so n = 3; the cumulative deformation, deformation rate, and marker density are all collected within a preset period.

[0087] It should be noted that the weights in this embodiment are determined using the coefficient of variation method. The coefficient of variation method assigns weights to each indicator based on the degree of variation between the current value and the target value of each evaluation indicator. If the numerical difference of an indicator is large, it can clearly distinguish each evaluated object, indicating that the indicator has rich discriminative information, and therefore should be given a larger weight. Conversely, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish each evaluated object is weak, and therefore should be given a smaller weight. This method directly utilizes the information contained in each indicator to calculate the weight of the indicator, and therefore has objectivity.

[0088] Results: By adopting a multi-source dataset fusion technology, a comprehensive and effective risk assessment of potential geological hazard points was achieved, resulting in high-precision risk identification. This not only improved the detection capability of potential landslide areas but also solved the problem of inaccurate risk assessment caused by a single data source in traditional methods.

[0089] S3. Landslide Occurrence Probability Prediction:

[0090] S3.1 GBDT Model Construction:

[0091] Using the GBDT model, the inputs are rainfall and temperature changes, and the output is the probability of a landslide occurring.

[0092] The GBDT model uses the Gradient Boosted Decision Tree (GBDT) algorithm to train a classification model based on historical landslide events and meteorological conditions.

[0093] When the GBDT model is running:

[0094] Input variables: cumulative rainfall P_rain (unit: mm) and temperature change (amplitude) T_change (unit: ℃) over a preset historical period; Output variable: probability of landslide occurrence P_landslide, where P_landslide ∈ [0, 1];

[0095] Example: Input the rainfall (more than 100 mm) and temperature change (average temperature below 10°C) over the past week, and the GBDT model predicts a 70% probability of landslides occurring in target area A.

[0096] S4. Correct the primary risk index to obtain the secondary risk index:

[0097] S4.1 Construct a risk index correction model to generate a secondary risk index R_sy for high-risk areas based on the probability of landslide occurrence. The method used is as follows:

[0098] ;

[0099] In the formula, α: correction coefficient, reflecting the strength of the impact of the landslide occurrence probability on the risk index;

[0100] Logical explanation: The higher the probability of a landslide, the greater the risk index should be, as the two are positively correlated. The logarithmic form makes the model more robust and avoids extreme value disturbances. Specifically, the logarithmic function is used to avoid the exponential explosion caused by the probability of a landslide approaching 1. The weight α is regionally adjusted according to the risk situation in the region, triggering the built-in parameter adjustment sub-model in the risk index correction model.

[0101] The suggested value for the correction factor α is greater than 0;

[0102] Results: The technical solution of predicting landslide probability using the GBDT model and correcting the primary risk index achieves dynamic adjustment from the primary risk index to the secondary risk index, resulting in more accurate risk quantification. This linkage mechanism not only considers geological conditions but also incorporates the impact of real-time meteorological changes, effectively solving the problem that traditional fixed-parameter models cannot adapt to complex environmental changes. For example, by analyzing rainfall and temperature changes, the probability of landslide occurrence can be updated in a timely manner, thereby adjusting the secondary risk index and ensuring the timeliness and accuracy of the early warning system.

[0103] In this embodiment, the parameter tuning sub-model operates by adjusting the correction coefficient α based on the number of high-risk areas in the entire target area and the primary risk index under each high-risk area. This parameter tuning method can more accurately reflect the overall geological hazard risk situation in the area. The operation process of the parameter tuning sub-model is as follows:

[0104] S4.1.1 Calculate the distribution characteristics of the number of high-risk areas and the primary risk index:

[0105] Collect the number N_r of all identified high-risk areas within the target area and their corresponding primary risk indices R_py. jWhere j = 1, 2, ..., N_r, j is a positive integer representing the number corresponding to the high-risk area; calculate the primary risk index R_py. j The mean R_py_avg and standard deviation σ_py;

[0106] Mean: ;

[0107] Standard deviation: ;

[0108] The adjustment formula for the correction factor α is determined as follows:

[0109] ;

[0110] In the formula, α0: initial weight, with a value range of [0.2, 0.4];

[0111] q1 and q2: Adjustment coefficients used to control the influence of the number of high-risk areas and the standard deviation of the primary risk index on α. ​​Their values ​​range from [0, 1], with recommended values ​​of 0.1 and 0.2 respectively. N_total: The number of grid cells or units in the target area. Therefore, the adjustment logic for the correction coefficient α is as follows:

[0112] Logic 1: If the number of high-risk areas N_r is large and the standard deviation σ_py of the primary risk index is large, it indicates that there are significant risk differences within the target area. The correction coefficient α needs to be increased to enhance the impact of the landslide probability on the secondary risk index and ensure that high-risk areas are given sufficient attention. Logic 2: Conversely, if the number of high-risk areas N_r or the standard deviation σ_py of the primary risk index is small, it indicates that the risk within the target area is relatively uniform. The correction coefficient α can be appropriately reduced to avoid over-amplifying the impact of local high-risk areas.

[0113] In summary, this approach allows for a better assessment of potential geological hazard risks within the region;

[0114] Effect description: The primary risk index plays a fundamental assessment role in geological disaster identification and early warning. It is a preliminary quantitative expression of the potential geological disaster risk in the region. It comprehensively considers key characteristics such as deformation rate, slope, and density of historical markers. By integrating these characteristics, an initial indicator reflecting the stability of high-risk areas is formed.

[0115] On the one hand, the primary risk index provides the basic input for risk modeling in subsequent models, enabling them to identify in depth whether there is high geological instability in high-risk areas, taking into account meteorological or cascading effects. On the other hand, the primary risk index can also serve as an important basis for dynamic parameter adjustment, guiding the adaptive adjustment of the correction coefficient α. By statistically analyzing the number of high-risk areas and their corresponding primary risk index distributions (such as mean and standard deviation), the degree of risk concentration and variability within the entire target area can be determined, thereby determining the weight of landslide probability in the calculation of the secondary risk index. This mechanism effectively solves the problem that fixed parameters in traditional methods cannot adapt to complex regional differences, realizing the transformation from "static assessment" to "dynamic response," and improving the sensitivity of the secondary risk index to local high-risk events and the robustness of the overall system.

[0116] S5. Comprehensive similarity comparison and risk level benchmarking:

[0117] S5.1, Comprehensive Similarity Calculation:

[0118] Compare the multi-source dataset corresponding to the secondary risk index R_sy in any high-risk area with similar data from the pre-disaster period of different historical geological (referring to landslide) disasters, and calculate the comprehensive similarity S_siy:

[0119] ;

[0120] In the formula, X i1 cu X is the i-th factor value in the current multi-source dataset (including at least cumulative deformation, deformation rate, slope, historical marker density, rainfall, and temperature). i1 hi : The i1th factor value in the historical multi-source dataset, i1: the corresponding weight coefficient, i1=1, 2, ..., n1, n1 represents the number of factor values ​​(i.e. factor dimension). If the factor values ​​include: cumulative deformation, deformation rate, slope, historical marker density, rainfall and temperature, then n1=6;

[0121] X i1 max and X i1 min : The historical maximum and minimum values ​​of this factor;

[0122] The final output S_sit∈[0,1] indicates that the larger the value, the more similar the two sides are. If they are completely identical, the value is 1; if they are completely different, the value approaches 0. This conversion method is intuitive and easy to set a threshold for subsequent judgment.

[0123] S5.2 Risk Level Benchmarking Mechanism:

[0124] Set the benchmark threshold S_th, for example: set S_th=0.75;

[0125] Iterate through all geological disaster event samples in the historical database, select several similar samples that satisfy S_siy≥S_th, and use the KNN algorithm or fuzzy membership function to perform level mapping, outputting the preliminary disaster risk level corresponding to the current high-risk area; among them, the existing geological disaster risk levels are divided into four levels: low, medium, high and extremely high. The rules for setting these levels are existing, so they will not be elaborated on here.

[0126] Additionally, it can output confidence levels and supports interpretations such as "medium level probability 80%";

[0127] The required warning levels for different risk levels can be found in Table 1:

[0128] Table 1: Comparison of Risk Levels and Warning Levels:

[0129]

[0130] As can be seen from Table 1, different risk levels are indicated by different warning levels;

[0131] Example: In this embodiment, the KNN algorithm is used for level mapping. Five samples with a similarity greater than 0.75 to the current high-risk area are found in the historical database; their disaster levels are: medium, medium, medium, high, and medium.

[0132] Therefore, the voting results are as follows: "Medium" received 4 votes and "High" received 1 vote → The current high-risk area is determined to be at the "Medium" level, so the preliminary disaster risk level is "Medium".

[0133] Results: By employing a technical solution combining comprehensive similarity calculation with the KNN algorithm or fuzzy membership function, intelligent benchmarking of current high-risk areas with historical disaster cases is achieved, realizing intelligent level mapping and improving the scientificity and reliability of disaster level determination. By comparing the similarity between current feature values ​​and historical disaster samples, the closest historical case can be quickly located, and the risk level of the current area can be determined accordingly, further enhancing the adaptive capability and response speed of the early warning system.

[0134] Unlike the subjectivity and limitations of traditional manual threshold setting (with the advantage of faster response time), and the improvements made to manual threshold setting in existing solutions by comprehensively calculating the required threshold based on historical data (with the advantage of higher accuracy), this method can effectively balance the accuracy of risk assessment and response speed, meeting practical needs.

[0135] S6. Analysis of the linkage effect between adjacent regions:

[0136] S6.1 Analysis of Regional Spacing and Geological Conditions:

[0137] Analyze the interval between two adjacent sub-risk indices that exceed the limit, and determine whether the interval and the two adjacent sub-risk indices that exceed the limit meet the set constraints. The constraints are as follows:

[0138] Condition 1: Are the geological conditions similar or identical?

[0139] Condition 2: Whether the maximum distance between the high-risk areas of the two sub-risk indices exceeding the standard is lower than the standard value;

[0140] Among them, for several high-risk areas with a historical risk level of "medium", their respective secondary risk indices are calculated, and the average value is taken as the standard indicator. If the secondary risk index exceeds the standard indicator, it is recorded as exceeding the standard.

[0141] When determining condition one, the following method can be used:

[0142] Acquire key layer data: including geological maps, slope maps (generated from DEM), and InSAR deformation maps; Extract target area attributes: for two adjacent high-risk areas and their intervals, extract their stratigraphic lithology (from geological maps), average slope (from DEM), and average annual deformation rate (from InSAR); Compare and analyze three indicators: Lithological consistency: if the three areas have the same lithological type (e.g., all are sandstone or shale), then the geological foundation is consistent; Slope difference: if the slope difference is < 5°, the topographic conditions are considered similar; Deformation rate similarity: if the rate difference is < 5 mm / yr, it indicates that the surface stability trend is consistent; Comprehensive judgment: if any two of the three indicators are consistent or highly similar, then the geological conditions of the interval area and the adjacent area can be determined to be the same or similar.

[0143] Note: The method described above is simple to operate and responds quickly. It is suitable for the rapid identification of spatially continuous areas in automated early warning methods. In practical applications, other methods can also be used to determine whether geological conditions are the same or similar. However, it should be noted that if the interval areas are discontinuous, it is directly determined that condition one is not met. The reason is that if the regions are discontinuous, a landslide in one high-risk area will have little or no impact on adjacent high-risk areas.

[0144] In the second judgment condition, the specific value set for the calibration value can be 1km, which can be set independently according to actual needs.

[0145] S6.2, Assessment of the linkage effect:

[0146] When the constraints are met, the linkage risk assessment mechanism is triggered (used to consider the impact of a high-risk area with a larger secondary risk index on another high-risk area, to assess whether there is a linkage effect, and if so, there may be an impact of expanding the risk level), to construct a linkage risk index model, input the geographical distance between the center points of the two excessive secondary risk indices and the high-risk areas corresponding to the two excessive secondary risk indices, and output the joint risk index R_cd;

[0147] The formula used when running the linked risk index model is as follows:

[0148] ;

[0149] In the formula, R_sy A and R_sy B : The secondary risk index exceeding the standard between two adjacent high-risk areas; β: the linkage enhancement coefficient, with a value range of [0, 1]; γ: the distance attenuation coefficient, which controls the rate at which the linkage effect weakens with distance, with a value range of [0, 0.1]; d: the geographical distance value;

[0150] Logical Explanation: This joint risk index formula scientifically quantifies the linkage effect of geological disasters between regions by comprehensively considering the secondary risk indices of two adjacent high-risk areas and their spatial distance relationship. The formula uses the larger risk value as the base term to reflect the influence of the dominant risk area. At the same time, it introduces the average risk and the linkage enhancement coefficient β to construct the enhancement term, and combines the exponential decay function to reflect the weakening effect of distance on the linkage effect. Thus, it realizes the dynamic assessment of the potential induced risks between adjacent areas. This method can not only identify the mutual influence between high-risk areas, but also adjust the linkage intensity according to the actual geographical distance. While improving the accuracy of the geological disaster early warning system, it also enhances the ability to predict and prevent regional chain disasters.

[0151] S6.3, Modify Disaster Level:

[0152] The joint risk index is compared with the standard index threshold. If the joint risk index exceeds the standard index threshold, it indicates that there is a disaster linkage effect between the two high-risk areas. The disaster level correction action is executed, the two adjacent high-risk areas are marked as a whole area, and the initial disaster risk level of the lower one is raised by one level to obtain the final disaster risk level. Based on the final disaster risk level, the corresponding warning level is sent according to Table 1.

[0153] For example, two adjacent high-risk areas that were originally classified as "medium" are marked as a whole area, resulting in a final disaster level of "high"; if there is one "medium" and one "high", then the "medium" is upgraded to "high", and the final disaster level for the whole area is "high".

[0154] If the joint risk index does not exceed the standard index threshold, it means that there is no disaster linkage effect between the two high-risk areas or the disaster linkage effect can be ignored. No response action is taken. The preliminary disaster risk level corresponding to the two adjacent high-risk areas is taken as the final disaster risk level. Table 1 should also be referred to here. Based on the final disaster risk level, the corresponding warning level should be sent.

[0155] For example, two adjacent high-risk areas that were originally classified as "medium" may both end up with a "medium" disaster level.

[0156] The standard index threshold is obtained in the same way as the standard indicators mentioned above, so it will not be elaborated on here. The standard index threshold is used to define whether the disaster linkage effect is serious enough to require a reassessment of the disaster risk level.

[0157] Results: Based on the joint risk index model and the technical solution for correcting disaster levels, this approach enables a quantitative assessment of potential chain reactions between adjacent high-risk areas, achieving a closed-loop effect for regional risk prevention and control. This solution addresses the shortcomings of traditional isolated analysis of high-risk points, which neglects the impact on surrounding areas, making the early warning system more comprehensive and flexible. Specifically, by calculating the joint risk index between two sub-risk indices exceeding the standard and adjusting the linkage intensity according to the actual geographical distance, it is possible to effectively predict and respond to possible chain disaster events, thereby changing the original risk level and sending corresponding early warnings to attract higher-level attention, thus improving the robustness and predictability of the overall solution.

[0158] S7. AI adaptive optimization mechanism based on data accumulation (which can be AI adaptive optimization performed under the condition of not triggering S6 and the analysis of the linkage effect between adjacent regions):

[0159] S7.1 Data Accumulation and Range Value Formation:

[0160] As data accumulates, the secondary risk index will form a range rather than a single value. Specifically, it will be generated using any one of the AI ​​models, such as GPR, BNN, or Random Forest, to determine the probability distribution range of the secondary risk index.

[0161] Example: The secondary risk index R_sy corresponding to a high-risk area is initially 40 → after accumulation, it becomes [38, 46];

[0162] S7.2, More Effective Benchmarking:

[0163] Regularly update the database used in the KNN algorithm and retrain the model with new samples to improve generalization ability;

[0164] Specifically, the application of AI technology is reflected in the following ways: Gaussian process regression (GPR) or Bayesian neural network (BNN) can be used to model the distribution characteristics of secondary risk indices; the model outputs a distribution with mean and variance based on input features (such as deformation rate, slope, rainfall, etc.); the final output is a confidence interval (e.g., 95% confidence interval [35, 45]), rather than a single value; as new data is added, the model continuously updates its parameters, causing the prediction interval to gradually converge and stabilize; through more historical data, the alignment between the secondary risk index and the known preliminary disaster risk level becomes more accurate; for example, assuming 100 similar cases have been accumulated, and it is found that when the secondary risk index is between 35 and 45, the disaster level is mostly "medium", then the risk level of the current area can be more accurately aligned, at which point S5.2 and the risk level alignment mechanism play a verification role; realizing the intelligent evolution from "static rule judgment" to "data-driven + AI-driven".

[0165] Example 2:

[0166] Based on Example 1, this embodiment also provides a geological disaster identification, monitoring and early warning system based on AI technology, which includes:

[0167] Data acquisition module: Collects multi-source datasets under the target area, preprocesses the multi-source datasets, and marks potential hazards to obtain several marker points; the multi-source datasets include at least: SAR data, UAV aerial imagery, geographic information, and meteorological data;

[0168] Risk identification module: Under the target area, based on geographic information, areas with a slope greater than 45° are identified as high-risk areas; based on multi-source datasets and hazard annotations, a feature set is selected under any high-risk area, and a primary risk index corresponding to each high-risk area is calculated based on the feature set;

[0169] Probability prediction module: Uses GBDT model, takes meteorological data as input and outputs the probability of landslide occurrence;

[0170] Risk Correction Module: Constructs a risk index correction model, corrects the primary risk index based on the probability of landslide occurrence, and derives the secondary risk index for the corresponding high-risk area;

[0171] Comprehensive benchmarking module: It compares the multi-source dataset corresponding to any secondary risk index with similar data from the previous period of different geological disasters in history, calculates the comprehensive similarity; sets a benchmarking threshold, iterates through all geological disaster event samples in the historical database, selects a number of similar samples, uses the KNN algorithm to perform level mapping, and outputs the preliminary disaster risk level;

[0172] Linkage Analysis Module: Analyzes the interval between two adjacent sub-risk indices that exceed the standard, determines whether the interval and the two adjacent sub-risk indices that exceed the standard meet the set constraints, and triggers the linkage risk assessment mechanism when they are met to generate a joint risk index; determines whether the joint risk index exceeds the standard index threshold.

[0173] If not, the preliminary disaster risk level will be used as the final disaster risk level;

[0174] If so, then execute the disaster level correction action and send the corresponding warning level based on the final disaster risk level;

[0175] Adaptive optimization module: Under the condition of continuous accumulation of the secondary risk index R_sy, the AI ​​model is used to generate the probability distribution interval of the secondary risk index R_sy; the database used in the KNN algorithm is updated regularly, and the AI ​​model is retrained with new samples; the AI ​​model is any one of GPR, BNN and random forest.

[0176] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0177] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0178] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for geological disaster identification, monitoring, and early warning based on AI technology, the method comprising: A multi-source dataset is collected from the target area, preprocessed, and hazard markers are added to obtain several marked points. The multi-source dataset includes at least: SAR data, UAV aerial imagery, geographic information, and meteorological data. The method further includes: Within the target area, areas with a slope greater than 45° are identified as high-risk areas based on geographic information. Based on multi-source datasets and hazard annotations, a feature set is selected for any high-risk area, and a primary risk index is calculated for each high-risk area based on the feature set. The feature set includes at least: cumulative deformation, deformation rate, and marker density. The primary risk index R_py for each high-risk area is calculated using the following method: ； In the formula, x i : The i-th eigenvalue, fi(): the standardization function, w i : The weights of each feature, i = 1, 2, ..., n, where n is the total number of features; The GBDT model is used, with meteorological data as input and landslide occurrence probability as output. A risk index correction model is constructed to correct the primary risk index based on the probability of landslide occurrence, thereby deriving the secondary risk index R_sy for the corresponding high-risk area. The method used is as follows: ； In the formula, α: correction coefficient, P_landslide: probability of landslide occurrence; Collect the number of all identified high-risk areas N_r within the target area and their corresponding primary risk index R_pyj, where j=1, 2, ..., N_r. Calculate the mean R_py_avg and standard deviation σ_py of the primary risk index R_pyj. The adjustment method for the correction factor α is as follows: ； In the formula, α0: initial weight; q1 and q2: adjustment coefficients, both ranging from [0, 1]; N_total: number of grids in the target region; The multi-source dataset corresponding to any secondary risk index is compared with similar data from the preceding periods of different historical geological disasters to calculate the comprehensive similarity. A benchmark threshold is set, and all geological disaster event samples in the historical database are traversed to select a number of similar samples. The KNN algorithm is used for level mapping to output the preliminary disaster risk level. Analyze the interval between two adjacent sub-risk indices that exceed the standard, determine whether the interval and the two adjacent sub-risk indices that exceed the standard meet the set constraints, and trigger the linkage risk assessment mechanism when they are met to generate a joint risk index; determine whether the joint risk index exceeds the standard index threshold. If not, the preliminary disaster risk level will be used as the final disaster risk level; If so, then execute the disaster level correction action and send the corresponding warning level based on the final disaster risk level; The process of triggering the joint risk assessment mechanism is as follows: Construct a joint risk index model, inputting two sub-risk indices that exceed the standard and the geographical distance between the center points of the high-risk areas corresponding to the two sub-risk indices, and outputting the joint risk index R_cd; The following methods are used when running the linked risk index model: ； In the formula, R_sy A and R_sy B : The secondary risk index exceeding the standard for two adjacent high-risk areas; β: the linkage enhancement coefficient, with a value range of [0, 1]; γ: the distance attenuation coefficient, with a value range of [0, 0.1]; d: the geographical distance value.

2. The method for geological disaster identification, monitoring, and early warning based on AI technology according to claim 1, characterized in that: The preprocessing of multi-source datasets includes at least the following: denoising, registration, and spatiotemporal alignment with meteorological data. SAR data should include at least: cumulative deformation and deformation rate; Drone aerial imagery includes at least: several images of the ground surface; Geographic information includes at least: topographic data and slope layers; Meteorological data should include at least: rainfall and temperature.

3. The method for geological disaster identification, monitoring, and early warning based on AI technology according to claim 1, characterized in that: The comprehensive similarity S_siy is calculated, and the conditions for selecting a number of similar samples are as follows: The overall similarity S_siy is greater than or equal to the benchmark threshold S_th.

4. The method for geological disaster identification, monitoring, and early warning based on AI technology according to claim 1, characterized in that: The constraints are: Condition 1: Are the geological conditions similar or identical? Condition 2: Whether the maximum distance between the high-risk areas of the two sub-risk indices that exceed the standard is lower than the standard value.

5. The method for geological disaster identification, monitoring, and early warning based on AI technology according to claim 1, characterized in that: The process of implementing disaster level correction is as follows: two adjacent high-risk areas are marked as a whole area, and the initial disaster risk level that is lower is raised by one level to obtain the final disaster risk level.

6. The method for geological disaster identification, monitoring, and early warning based on AI technology according to claim 1, characterized in that: The method also includes: generating the probability distribution interval of the secondary risk index R_sy using an AI model under the condition that the secondary risk index R_sy continues to accumulate; periodically updating the database used in the KNN algorithm and retraining the AI ​​model using new samples; wherein the AI ​​model is any one of GPR, BNN and random forest.

7. A geological disaster identification, monitoring, and early warning system based on AI technology, the system comprising: Data acquisition module: Collects multi-source datasets from the target area, preprocesses the multi-source datasets, and marks potential hazards to obtain several marker points; wherein, the multi-source datasets include at least: SAR data, UAV aerial imagery, geographic information, and meteorological data; characterized in that: the system also includes: Risk identification module: Under the target area, based on geographic information, identify areas with a slope greater than 45° as high-risk areas; based on multi-source datasets and hazard annotations, select feature sets for any high-risk area, and calculate the primary risk index corresponding to each high-risk area based on the feature sets; the feature set includes at least: cumulative deformation, deformation rate, and marker density; The primary risk index R_py for each high-risk area is calculated using the following method: ； In the formula, x i : The i-th eigenvalue, fi(): the standardization function, w i : The weights of each feature, i = 1, 2, ..., n, where n is the total number of features; Probability prediction module: Uses GBDT model, takes meteorological data as input and outputs the probability of landslide occurrence; Risk Correction Module: Constructs a risk index correction model, corrects the primary risk index based on the probability of landslide occurrence, and derives the secondary risk index R_sy for the corresponding high-risk area. The method used is as follows: ； In the formula, α: correction coefficient, P_landslide: probability of landslide occurrence; Collect the number of all identified high-risk areas N_r within the target area and their corresponding primary risk index R_pyj, where j=1, 2, ..., N_r. Calculate the mean R_py_avg and standard deviation σ_py of the primary risk index R_pyj. The adjustment method for the correction factor α is as follows: ； In the formula, α0: initial weight; q1 and q2: adjustment coefficients, both ranging from [0, 1]; N_total: number of grids in the target region; Comprehensive benchmarking module: It compares the multi-source dataset corresponding to any secondary risk index with similar data from the previous period of different geological disasters in history, calculates the comprehensive similarity; sets a benchmarking threshold, iterates through all geological disaster event samples in the historical database, selects a number of similar samples, uses the KNN algorithm to perform level mapping, and outputs the preliminary disaster risk level; Linkage Analysis Module: Analyzes the interval between two adjacent sub-risk indices that exceed the standard, determines whether the interval and the two adjacent sub-risk indices that exceed the standard meet the set constraints, and triggers the linkage risk assessment mechanism when they are met to generate a joint risk index; determines whether the joint risk index exceeds the standard index threshold. If not, the preliminary disaster risk level will be used as the final disaster risk level; If so, then execute the disaster level correction action and send the corresponding warning level based on the final disaster risk level; The process of triggering the joint risk assessment mechanism is as follows: Construct a joint risk index model, inputting two sub-risk indices that exceed the standard and the geographical distance between the center points of the high-risk areas corresponding to the two sub-risk indices, and outputting the joint risk index R_cd; The following methods are used when running the linked risk index model: ； In the formula, R_sy A and R_sy B : The secondary risk index exceeding the standard between two adjacent high-risk areas; β: the linkage enhancement coefficient, with a value range of [0, 1]; γ: the distance attenuation coefficient, with a value range of [0, 0.1]; d: the geographical distance value; Adaptive optimization module: Under the condition of continuous accumulation of the secondary risk index R_sy, the AI ​​model is used to generate the probability distribution interval of the secondary risk index R_sy; the database used in the KNN algorithm is updated regularly, and the AI ​​model is retrained with new samples; the AI ​​model is any one of GPR, BNN and random forest.

Images