Big data-based geological disaster risk assessment platform and method

Abstract

The invention relates to the technical field of big data analysis, in particular to a geological disaster risk assessment platform and method based on big data, and the platform comprises a multi-source data collection module, a dynamic weight fusion module, a dual-channel time sequence modeling module, a risk probability dynamic deduction module, a multi-dimensional coupling assessment module and a three-dimensional visualization early warning module. Wherein the multi-source data acquisition module is used for acquiring multi-source data in real time; the dynamic weight fusion module is used for generating a fusion data set of time-space alignment; the dual-channel time sequence modeling module is used for outputting a coupling feature vector; and the risk probability dynamic deduction module is used for generating a regional geological risk probability distribution diagram. According to the geological disaster risk assessment system and method, accurate geological disaster risk prediction and assessment are provided by fusing multi-source data and adopting an advanced time sequence analysis and assessment technology, an interactive risk map can be generated in real time, graded early warning signals can be triggered, and data support is provided for disaster management and emergency response.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and particularly to a geological disaster risk assessment platform and method based on big data. Background Art

[0002] Geological disasters are characterized by suddenness, strong destructiveness and wide influence range. Traditional geological disaster assessment and early warning methods often rely on limited historical data and on-site monitoring, making it difficult to achieve real-time and accurate disaster risk prediction. With the rapid development of remote sensing technology, sensor networks and big data technology, there are gradually more comprehensive data sources, which provide new technical means for the assessment and prediction of geological disasters. However, how to extract valuable information from massive and multi-source data and conduct efficient comprehensive analysis remains an urgent problem to be solved.

[0003] Existing geological disaster risk assessment technologies have many limitations, mainly reflected in the insufficient data processing and analysis capabilities. Traditional methods mostly rely on a single data source, lacking effective fusion and analysis of multi-dimensional and multi-source data, resulting in poor accuracy and timeliness of disaster prediction. In addition, existing models are difficult to comprehensively consider the influence of complex factors such as geological structures and human activities on geological disaster risks, and the assessment results are often inaccurate and cannot reflect the disaster risk situations of each region in real time. Summary of the Invention

[0004] Based on the above purposes, the present invention provides a geological disaster risk assessment platform and method based on big data.

[0005] The geological disaster risk assessment platform based on big data includes a multi-source data acquisition module, a dynamic weight fusion module, a dual-channel time series modeling module, a risk probability dynamic deduction module, a multi-dimensional coupling assessment module and a three-dimensional visualization early warning module; wherein:

[0006] The multi-source data acquisition module: is used to obtain satellite remote sensing data, groundwater level monitoring data, surface displacement sensor data and meteorological data in real time;

[0007] The dynamic weight fusion module: is used to receive the output data of the multi-source data acquisition module and generate a spatio-temporally aligned fusion data set by using an adaptive feature selection algorithm;

[0008] The dual-channel time series modeling module: is used to receive the fusion data set and output a coupled feature vector through the parallel processing of a surface deformation feature extraction channel and a groundwater hydrological anomaly detection channel;

[0009] The risk probability dynamic deduction module: is used to input the coupled feature vector into an LSTM-Transformer hybrid model optimized by transfer learning to generate a regional geological risk probability distribution map;

[0010] Multi - dimensional Coupling Evaluation Module: It is used to generate an evaluation matrix containing stability index, interference sensitivity, and catastrophe diffusion rate according to the risk probability distribution map, combined with the geological structure feature library and the human activity database;

[0011] Three - dimensional Visualization Early Warning Module: It is used to receive the evaluation matrix data, generate an interactive risk map through dynamic isosurface rendering technology, and trigger a hierarchical early warning signal.

[0012] Optionally, the multi - source data acquisition module includes a satellite remote sensing data acquisition unit, a groundwater level monitoring data acquisition unit, a surface displacement sensor data acquisition unit, and a meteorological data acquisition unit; among them:

[0013] Satellite Remote Sensing Data Acquisition Unit: It uses remote sensors deployed on satellites to collect spectral data of the ground area in real - time, including data in visible light, infrared, and microwave bands;

[0014] Groundwater Level Monitoring Data Acquisition Unit: It uses water level monitoring sensors installed in groundwater sources to monitor the changes in groundwater level in real - time;

[0015] Surface Displacement Sensor Data Acquisition Unit: It uses displacement sensors installed on the ground to monitor the surface displacement in real - time, and records the direction and amplitude of the ground displacement;

[0016] Meteorological Data Acquisition Unit: It is used to collect meteorological data in real - time, including temperature, humidity, wind speed, and precipitation.

[0017] Optionally, the dynamic weight fusion module includes a data pre - processing unit, an adaptive feature selection unit, and a spatio - temporal alignment unit; among them:

[0018] Data Pre - processing Unit: It is used to pre - process the data output by the multi - source data acquisition module, including data denoising, normalization, and missing value filling, to ensure that the data formats of different data sources are consistent;

[0019] Adaptive Feature Selection Unit: It analyzes the pre - processed data through an adaptive feature selection algorithm, automatically identifies and selects the most relevant feature variables, and excludes redundant information;

[0020] Spatio - temporal Alignment Unit: It is used to align the feature data from different data sources in terms of time and space, ensure the matching of timestamps and geographical locations of each data source, so as to generate a spatio - temporally aligned fusion data set.

[0021] Optionally, the adaptive feature selection unit includes:

[0022] Feature Importance Calculation: The Pearson correlation coefficient is used to measure the linear relationship between each feature and the target variable;

[0023] Redundant feature elimination: For the redundant features in the feature set, by calculating the correlation between features, those features that are highly correlated with other features are removed. Specifically, when the correlation between a certain feature and other features exceeds a predetermined threshold, the feature exceeding the predetermined threshold is considered a redundant feature and is thus eliminated.

[0024] Optionally, the dual-channel time series modeling module includes a surface deformation feature extraction channel, a groundwater hydrological anomaly detection channel, and a coupled feature vector output unit; where:

[0025] Surface deformation feature extraction channel: Used to receive surface displacement data from the multi-source data acquisition module and extract the features of surface deformation through time series modeling technology;

[0026] Groundwater hydrological anomaly detection channel: Used to receive groundwater level data and extract groundwater hydrological anomaly features through time series analysis technology;

[0027] Coupled feature vector output unit: Used to perform parallel processing on the output results of the surface deformation feature extraction channel and the groundwater hydrological anomaly detection channel, and fuse the feature vectors of the two channels in series to generate a comprehensive coupled feature vector.

[0028] Optionally, the coupled feature vector output unit includes:

[0029] Feature vector standardization: Perform standardization processing on the surface deformation feature vector and the groundwater hydrological anomaly feature vector. Let the surface deformation feature vector be F surface , and the groundwater hydrological anomaly feature vector be F water , then the standardization formula is: and where, μ surface and σ surface are respectively the mean and standard deviation of the surface deformation feature vector, μ water and σ water are respectively the mean and standard deviation of the groundwater hydrological anomaly feature vector, F surface ′ and F water ′ are the standardized feature vectors;

[0030] Feature vector concatenation: Fuse the standardized surface deformation feature vector and the groundwater hydrological anomaly feature vector through concatenation operation. The concatenation operation arranges the two feature vectors in sequence together to form a comprehensive coupled feature vector, expressed as: F coupled = [F surface ′; F water ′], where, F coupled is the output comprehensive coupled feature vector, F surface ′ and Fwater ′ is the standardized surface deformation feature vector and groundwater hydrological anomaly feature vector, and the symbol ; represents the concatenation operation of vectors.

[0031] Optionally, the risk probability dynamic deduction module includes a coupled feature vector input unit, an LSTM unit, a Transformer unit, and a geological risk probability distribution map generation unit; where:

[0032] Coupled feature vector input unit: used to receive the coupled feature vector F output from the dual-channel time series modeling module, and use it as input data to be passed to the LSTM-Transformer hybrid model for analysis; coupled , and pass it as input data to the LSTM-Transformer hybrid model for analysis;

[0033] LSTM unit: used to capture the time series pattern in the coupled feature vector, selectively save important historical information through the memory mechanism, and ignore irrelevant parts. Its calculation formula is: h t =σ(W h x t +b h )·tanh(W c c t-1 +b c ), where h t is the hidden state at the current moment, x t is the input at the current moment, c t-1 is the cell state at the previous moment, W h and W c are weight matrices, b h and b c are bias terms;

[0034] Transformer unit: used to capture the long-term dependencies in the coupled feature vector and enhance the parallel processing ability of the model. Transformer processes the relationships between various parts of the input data through the self-attention mechanism to capture the global information in the input data. The calculation formula of the self-attention mechanism is:

[0035] Among them, Q, K, and V are the query, key, and value respectively, and d k is the dimension of the key;

[0036] Geological risk probability distribution map generation unit: used to convert the output data processed by the LSTM-Transformer hybrid model into a geological risk probability distribution map.

[0037] Optionally, the multi-dimensional coupling evaluation module includes a geological structure feature extraction unit, a human activity impact assessment unit, a disaster diffusion rate calculation unit, and an evaluation matrix generation unit; where:

[0038] Geological structure feature extraction unit: used to extract relevant data from the geological structure feature library and calculate the geological stability of each region. The formula is: where SI(x, y) is the stability index at position (x, y), and G i (x, y) is the value of the i-th geological feature at position (x, y), and w i is the weight of this geological feature, and n1 is the total number of geological features;

[0039] Human activity impact assessment unit: used to extract activity data related to geological disaster risks from the human activity database, including population density, urbanization level, and infrastructure construction factors; and calculate the interference sensitivity based on these relevant activity data. The formula is: where IS(x, y) is the interference sensitivity at position (x, y), and A i (x, y) is the value of the i-th human activity at position (x, y), and α i is the impact factor of this human activity, and m is the number of human activity factors;

[0040] Catastrophe diffusion rate calculation unit: used to calculate the catastrophe diffusion rate by analyzing the propagation mode of geological disasters in the region. The formula is: where DR(x, y) is the catastrophe diffusion rate at position (x, y), P(x, y) is the geological disaster risk probability at this position, d(x, y) is the geological disaster propagation distance at this position, and k is the diffusion coefficient;

[0041] Evaluation matrix generation unit: used to integrate the data extracted from the geological structure feature library and the human activity database, as well as the stability index, interference sensitivity, and catastrophe diffusion rate calculated from the above units, to generate an evaluation matrix containing these three indicators.

[0042] Optionally, the three-dimensional visualization and early warning module includes a dynamic isosurface rendering unit, an interactive atlas generation unit, and a graded early warning signal triggering unit; where:

[0043] Dynamic isosurface rendering unit: uses dynamic isosurface rendering technology to generate a three-dimensional risk atlas based on the risk data in the evaluation matrix. Specifically, first, determine the corresponding risk level according to the stability index, interference sensitivity, and catastrophe diffusion rate of each region, and map the risk level to different color or transparency values; then use the isosurface algorithm to generate an isosurface representing the risk degree and display the risk distribution of different regions;

[0044] Interactive atlas generation unit: used to generate an interactive three-dimensional model from the rendered risk atlas, allowing users to dynamically view the risk distribution of different regions through rotation and zoom operations;

[0045] Hierarchical warning signal trigger unit: used to trigger corresponding hierarchical warning signals according to preset risk levels. The basis for dividing warning levels is expressed as:

[0046]

[0047] where threshold2 and threshold3 are respectively the preset low-risk and high-risk thresholds, P(x,y) is the risk probability at this location, and Alert Level is the triggered warning level.

[0048] The geological disaster risk assessment method based on big data is implemented by the above-mentioned geological disaster risk assessment platform based on big data, and includes the following steps:

[0049] S1: Real-time collect satellite remote sensing data, groundwater level monitoring data, surface displacement data and meteorological data of the ground area;

[0050] S2: Preprocess the collected various data, including data denoising, normalization and missing value filling;

[0051] S3: Analyze the processed data through an adaptive feature selection algorithm, identify and select the most relevant feature variables, and exclude redundant information;

[0052] S4: Based on the feature data selected in S3, extract the time series features of surface deformation and groundwater hydrological anomalies, and calculate the coupled feature vector;

[0053] S5: Input the coupled feature vector into the LSTM-Transformer hybrid model for time series analysis and prediction, and generate a probability distribution map of regional geological risks;

[0054] S6: According to the geological risk probability distribution map, combined with geological structure characteristics and human activity data, calculate the stability index, interference sensitivity and disaster diffusion rate of each region, and generate an evaluation matrix;

[0055] S7: According to the evaluation matrix, generate a three-dimensional risk map, use dynamic isosurface rendering technology to visualize the geological disaster risks of different regions, and trigger corresponding warning signals according to the risk levels.

[0056] Advantages of the present invention:

[0057] In the present invention, by collecting multi-source information such as satellite remote sensing data, groundwater level monitoring data, surface displacement data, and meteorological data, and using an adaptive feature selection algorithm to process and analyze the data, efficient data fusion is achieved. This technology can effectively remove redundant information, retain key features related to geological disaster risks, and significantly improve the accuracy and efficiency of data processing. By introducing an LSTM-Transformer hybrid model for time series analysis, the spatio-temporal evolution law of geological disasters can be accurately captured, providing a more reliable basis for disaster risk prediction and regional assessment.

[0058] In the present invention, by combining geological structure features and human activity data, evaluation indexes such as stability index, interference sensitivity, and catastrophe diffusion rate are generated to comprehensively evaluate the disaster risks of each region, and an interactive risk map is generated through dynamic isosurface rendering technology. This visualization technology not only improves the intuitiveness of risk assessment but also can trigger graded warning signals in real time, providing strong support for disaster emergency response and resource allocation. Brief Description of the Drawings

[0059] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only those of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0060] Figure 1 Schematic diagram of the geological disaster risk assessment platform according to an embodiment of the present invention;

[0061] Figure 2 Schematic diagram of the risk probability dynamic deduction module according to an embodiment of the present invention. Detailed Embodiments

[0062] The present invention will be described in detail below with reference to the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; moreover, the drawing part is only for more specifically describing the embodiments and is not intended to specifically limit the present invention.

[0063] It should be noted that in the specification, references to "an embodiment", "embodiments", "exemplary embodiments", "some embodiments", etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment includes such specific features, structures, or characteristics. Additionally, when combining embodiments to describe a specific feature, structure, or characteristic, implementing such feature, structure, or characteristic in combination with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the relevant art.

[0064] Generally, terms can be understood, at least in part, from their use in context. For example, at least in part depending on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or can be used to describe a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey a set of exclusive factors, but rather, at least in part depending on the context, can allow for the existence of other factors that are not necessarily explicitly described.

[0065] As Figure 1 shown, a geological disaster risk assessment platform based on big data includes a multi-source data acquisition module, a dynamic weight fusion module, a dual-channel time series modeling module, a risk probability dynamic deduction module, a multi-dimensional coupling evaluation module, and a three-dimensional visualization warning module; wherein:

[0066] Multi-source data acquisition module: used to obtain satellite remote sensing data, groundwater level monitoring data, surface displacement sensor data, and meteorological data in real time;

[0067] Dynamic weight fusion module: used to receive the output data of the multi-source data acquisition module and generate a spatio-temporally aligned fusion data set using an adaptive feature selection algorithm;

[0068] Dual-channel time series modeling module: used to receive the fusion data set and output a coupled feature vector through the parallel processing of the surface deformation feature extraction channel and the groundwater hydrological anomaly detection channel;

[0069] Risk probability dynamic deduction module: used to input the coupled feature vector into an LSTM-Transformer hybrid model optimized by transfer learning to generate a regional geological risk probability distribution map;

[0070] Multi-dimensional coupling evaluation module: used to generate an evaluation matrix including a stability index, interference sensitivity, and disaster diffusion rate according to the risk probability distribution map, in combination with a geological structure feature library and a human activity database;

[0071] Three-dimensional visualization warning module: used to receive the evaluation matrix data, generate an interactive risk map through dynamic isosurface rendering technology, and trigger a graded warning signal.

[0072] The multi-source data acquisition module includes a satellite remote sensing data acquisition unit, a groundwater level monitoring data acquisition unit, a surface displacement sensor data acquisition unit, and a meteorological data acquisition unit; among them:

[0073] Satellite remote sensing data acquisition unit: Through remote sensing sensors deployed on satellites, it collects spectral data of the ground area in real time, including data in visible light, infrared, and microwave bands, and monitors surface temperature, vegetation coverage, and humidity indicators;

[0074] Groundwater level monitoring data acquisition unit: Through water level monitoring sensors installed in underground water sources, it monitors the change of groundwater level in real time;

[0075] Surface displacement sensor data acquisition unit: Through displacement sensors installed on the ground, it monitors the surface displacement in real time and records the direction and amplitude of the ground displacement;

[0076] Meteorological data acquisition unit: Used to collect meteorological data in real time, including air temperature, humidity, wind speed, and precipitation; Through the collaborative work of the above multiple units, the multi-source data acquisition module can obtain key data from different sources simultaneously and in real time, and transmit it to the subsequent processing module, providing accurate input data for geological disaster risk assessment.

[0077] The dynamic weight fusion module includes a data preprocessing unit, an adaptive feature selection unit, and a spatio-temporal alignment unit; among them:

[0078] Data preprocessing unit: Used to preprocess the data output by the multi-source data acquisition module, including data denoising, normalization, and missing value filling, ensuring that the data formats of different data sources are consistent and meet the requirements of subsequent processing;

[0079] Adaptive feature selection unit: Analyzes the preprocessed data through an adaptive feature selection algorithm, automatically identifies and selects the most relevant feature variables, and excludes redundant information, thereby improving the efficiency and accuracy of data processing. This algorithm adjusts the feature selection according to the weights of each data source to generate a more accurate spatio-temporal alignment data set;

[0080] Spatio-temporal alignment unit: Used to align the feature data from different data sources in time and space, ensuring the matching of timestamps and geographical locations of each data source, thereby generating a spatio-temporal aligned fusion data set for subsequent modeling and analysis.

[0081] The adaptive feature selection unit includes:

[0082] Feature importance calculation: The Pearson correlation coefficient is used to measure the linear relationship between each feature and the target variable. The formula is: Among them, X kiis the value of the k-th feature in the i-th sample, Y li is the value of the target variable in the i-th sample, and are the means of the k-th feature and the l-th target variable respectively, and n is the number of samples; the correlation coefficient calculated by this formula can help identify which feature variables are most closely related to the target variable;

[0083] Redundant feature elimination: For the redundant features in the feature set, by calculating the correlation between features (using Pearson correlation coefficient or other correlation measurement methods), remove those features that are highly correlated with other features. Specifically, when the correlation of a certain feature with other features exceeds a certain predetermined threshold (such as 0.95), the feature exceeding the predetermined threshold is considered a redundant feature and is then eliminated; through the above steps, finally, a feature set that is most relevant to geological hazard risk assessment and has the least redundant information is selected.

[0084] The dual-channel time series modeling module includes a surface deformation feature extraction channel, a groundwater hydrological anomaly detection channel, and a coupled feature vector output unit; among them:

[0085] Surface deformation feature extraction channel: Used to receive surface displacement data from the multi-source data acquisition module, and extract the features of surface deformation through time series modeling technology; specifically, first, by performing differential processing on the surface displacement data, calculate the change trend of surface deformation; then, use the autoregressive model (AR) to perform time series modeling on the surface displacement data, extract the time series features of surface deformation, and output the deformation feature vector through normalization processing;

[0086] Groundwater hydrological anomaly detection channel: Used to receive groundwater level data and extract groundwater hydrological anomaly features through time series analysis technology; specifically, first, remove the noise in the groundwater level data through a filtering algorithm to ensure the accuracy of the data; then, use an anomaly detection algorithm (such as Isolation Forest) to detect the abnormal changes in the groundwater level and identify possible hydrological anomaly events;

[0087] Coupled feature vector output unit: Used to perform parallel processing on the output results of the surface deformation feature extraction channel and the groundwater hydrological anomaly detection channel, and fuse the feature vectors of the two channels in series to generate a comprehensive coupled feature vector. This coupled feature vector contains surface deformation and groundwater hydrological anomaly information, providing comprehensive input data for subsequent risk assessment and prediction; through the above units, the dual-channel time series modeling module can effectively generate a comprehensive coupled feature vector by parallel processing surface deformation and groundwater hydrological anomaly features, and then provide comprehensive feature data support for geological hazard risk assessment.

[0088] The coupled feature vector output unit includes:

[0089] Feature vector normalization: Normalize the surface deformation feature vector and the groundwater hydrological anomaly feature vector to ensure that their numerical ranges are the same. Let the surface deformation feature vector be F surface , and the groundwater hydrological anomaly feature vector be F water . Then the normalization formula is: and where μ surface and σ surface are the mean and standard deviation of the surface deformation feature vector respectively, μ water and σ water are the mean and standard deviation of the groundwater hydrological anomaly feature vector respectively, F surface ′ and F water ′ are the normalized feature vectors;

[0090] Feature vector concatenation: Concatenate the normalized surface deformation feature vector and the groundwater hydrological anomaly feature vector through a concatenation operation for fusion. The concatenation operation arranges the two feature vectors in sequence to form a comprehensive coupled feature vector, denoted as: F coupled =[F surface ′; F water ′], where F coupled is the output comprehensive coupled feature vector, F surface ′ and F water ′ are the normalized surface deformation feature vector and the groundwater hydrological anomaly feature vector respectively, and the symbol; represents the concatenation operation of vectors; Through the above steps, the coupled feature vector output unit can effectively fuse the feature vectors from the two channels through normalization, concatenation, and weighting, etc., to generate a comprehensive coupled feature vector, providing accurate and comprehensive input features for subsequent geological disaster risk assessment.

[0091] The risk probability dynamic deduction module includes a coupled feature vector input unit, an LSTM unit, a Transformer unit, and a geological risk probability distribution map generation unit; among them:

[0092] Coupled feature vector input unit: Used to receive the coupled feature vector F coupled output from the dual-channel time series modeling module, and use it as input data to be passed to the LSTM-Transformer hybrid model for analysis. This unit ensures that the feature vector is input in the correct format, guarantees data consistency and integrity, and provides accurate feature input for subsequent analysis;

[0093] LSTM unit: It is used to capture the time series patterns in the coupled feature vectors, selectively save important historical information through the memory mechanism, and ignore the irrelevant parts. Its calculation formula is: h t = σ(W h x t + b h ) · tanh(W c c t-1 + b c ), where h t is the hidden state at the current moment, x t is the input at the current moment, c t-1 is the cell state at the previous moment, W h and W c are weight matrices, b h and b c are bias terms. Through this calculation, the LSTM unit can extract the time series patterns in the coupled feature vectors and output the time series features;

[0094] Transformer unit: It is used to capture the long-term dependencies in the coupled feature vectors and enhance the parallel processing ability of the model. The Transformer processes the relationships between the various parts of the input data through the self-attention mechanism (Self-Attention) to capture the global information in the input data. The calculation formula of the self-attention mechanism is: where Q, K, and V are the query, key, and value respectively, and d k is the dimension of the key. Through this formula, the Transformer unit can assign different attention weights to different parts of the input data, thereby effectively capturing the global relationships of the input data;

[0095] Geological risk probability distribution map generation unit: It is used to convert the output data processed by the LSTM-Transformer hybrid model into a geological risk probability distribution map. This process is carried out through the following formula:

[0096] where P(x, y) is the geological disaster risk probability at the location (x, y), z xy is the feature vector at the location (x, y), and w is the weight parameter of the model. Through this formula calculation, the risk probability value of each region is generated and visualized as a geological risk distribution map; Through the above-mentioned units, the risk probability dynamic deduction module can analyze the coupled feature vectors based on the LSTM-Transformer hybrid model, generate the regional geological risk probability distribution map, and provide decision support for subsequent geological disaster prediction and emergency response.

[0097] The multi-dimensional coupling evaluation module includes a geological structure feature extraction unit, a human activity impact assessment unit, a disaster diffusion rate calculation unit, and an evaluation matrix generation unit; among which:

[0098] Geological structure feature extraction unit: It is used to extract relevant data from the geological structure feature library and calculate the geological stability of each region. The geological structure feature library includes the rock layer structure, fault distribution, and seismic activity frequency information of each region. The stability index of each region is calculated by the weighted average method, and the formula is: Among them, SI(x, y) is the stability index of position (x, y), and G i (x, y) is the value of the i-th geological feature (such as rock layer type, fault activity, etc.) at position (x, y), and w i is the weight of this geological feature, and n1 is the total number of geological features. Through this process, a stability index can be assigned to each region to reflect the stability of its geological structure;

[0099] Human activity impact assessment unit: It is used to extract activity data related to geological disaster risks from the human activity database, including population density, urbanization level, and infrastructure construction factors; and calculate the interference sensitivity according to these relevant activity data, and the formula is: Among them, IS(x, y) is the interference sensitivity of position (x, y), and A i (x, y) is the value of the i-th human activity (such as building density, traffic flow, etc.) at position (x, y), and α i is the impact factor of this human activity, and m is the number of human activity factors. This process calculates the interference sensitivity for each region to quantify the potential impact of human activities on geological disaster risks;

[0100] Disaster diffusion rate calculation unit: It is used to calculate the disaster diffusion rate by analyzing the propagation mode of geological disasters in the region. Through a calculation method based on the diffusion model, according to factors such as the geological characteristics, meteorological conditions, and terrain changes of the region, the diffusion rate of disasters in space is deduced, and the formula is: Among them, DR(x, y) is the disaster diffusion rate of position (x, y), P(x, y) is the geological disaster risk probability of this position, d(x, y) is the geological disaster propagation distance of this position, and k is the diffusion coefficient. Through this calculation, a disaster diffusion rate value can be assigned to each region to reflect the spread speed of disasters;

[0101] Evaluation matrix generation unit: It is used to synthesize the data extracted from the geological structure feature library and the human activity database, as well as the stability index, interference sensitivity, and disaster diffusion rate calculated from the above units, to generate an evaluation matrix EM(x, y) that includes these three indicators; this evaluation matrix provides a comprehensive analysis of the geological disaster risk for each region, and the expression is:

[0102] EM(x, y) = [SI(x, y) IS(x, y) DR(x, y)], where EM(x, y) is the evaluation matrix at position (x, y), including the stability index SI(x, y), interference sensitivity IS(x, y), and disaster diffusion rate DR(x, y); through the above units, the multi-dimensional coupling evaluation module can combine the geological structure feature library and the human activity database to generate an evaluation matrix that includes the stability index, interference sensitivity, and disaster diffusion rate, providing comprehensive data support for subsequent risk assessment and disaster warning.

[0103] The three-dimensional visualization warning module includes a dynamic isosurface rendering unit, an interactive atlas generation unit, and a graded warning signal triggering unit; among them:

[0104] Dynamic isosurface rendering unit: Using dynamic isosurface rendering technology, it generates a three-dimensional risk atlas based on the risk data in the evaluation matrix. Specifically, first, according to the stability index, interference sensitivity, and disaster diffusion rate of each region, determine its corresponding risk level, and map the risk level to different color or transparency values; then use the isosurface algorithm to generate an isosurface representing the risk degree to display the risk distribution of different regions; the isosurface calculation formula is as follows: where f(x, y, z) is the risk level value at position (x, y, z), P(x, y) is the risk probability at this position, and threshold1 is a preset risk threshold; through this formula, the dynamic isosurface rendering unit can generate a three-dimensional visual risk atlas based on the risk data;

[0105] Interactive atlas generation unit: It is used to generate an interactive three-dimensional model through the rendered risk atlas, allowing users to dynamically view the risk distribution of different regions through rotation and zoom operations. Users can select specific regions or risk levels to further view the detailed risk data of that region, such as the stability index, interference sensitivity, etc., enhancing the visualization effect and user experience of risk assessment;

[0106] Graded warning signal triggering unit: It is used to trigger corresponding graded warning signals according to the preset risk levels. The basis for dividing the warning levels is expressed as:

[0107]

[0108] Among them, threshold2 and threshold3 are respectively the preset low-risk and high-risk thresholds, P(x,y) is the risk probability at this location, Alert Level is the triggered warning level, and according to different risk levels, high, medium, and low-level warning signals are triggered to notify relevant departments or personnel; through the above unit, the 3D visualization warning module can generate an interactive risk map based on the dynamic isosurface rendering technology and trigger graded warning signals according to the risk data, thus providing visual support and decision-making basis for the real-time monitoring and emergency response of geological disasters.

[0109] As Figure 2 shown, the geological disaster risk assessment method based on big data is implemented by the above-mentioned geological disaster risk assessment platform based on big data, and includes the following steps:

[0110] S1: Real-time collect satellite remote sensing data, groundwater level monitoring data, surface displacement data, and meteorological data of the ground area;

[0111] S2: Preprocess the collected various data, including data denoising, normalization, and missing value filling, to ensure data consistency;

[0112] S3: Analyze the processed data through an adaptive feature selection algorithm, identify and select the most relevant feature variables, and exclude redundant information;

[0113] S4: Based on the feature data selected in S3, extract the time series features of surface deformation and groundwater hydrological anomalies, and calculate the coupled feature vector;

[0114] S5: Input the coupled feature vector into the LSTM-Transformer hybrid model for time series analysis and prediction, and generate a probability distribution map of regional geological risks;

[0115] S6: According to the geological risk probability distribution map, combined with geological structure features and human activity data, calculate the stability index, interference sensitivity, and disaster diffusion rate of each region, and generate an evaluation matrix;

[0116] S7: According to the evaluation matrix, generate a 3D risk map, use the dynamic isosurface rendering technology to visualize the geological disaster risks of different regions, and trigger corresponding warning signals according to the risk levels.

[0117] The present invention encompasses any alternatives, modifications, equivalent methods, and solutions made to the essence and scope of the present invention. To enable the public to thoroughly understand the present invention, specific details are described in detail in the following preferred embodiments of the present invention, and those skilled in the art can fully understand the present invention without the description of these details. Additionally, well-known methods, processes, procedures, components, and circuits, etc. are not described in detail to avoid unnecessary confusion to the essence of the present invention.

[0118] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A geological disaster risk assessment platform based on big data, characterized in that It includes a multi-source data acquisition module, a dynamic weight fusion module, a dual-channel time series modeling module, a risk probability dynamic deduction module, a multi-dimensional coupling evaluation module, and a three-dimensional visualization warning module; among them: The multi-source data acquisition module: is used to obtain satellite remote sensing data, groundwater level monitoring data, surface displacement sensor data, and meteorological data in real time; The dynamic weight fusion module: is used to receive the output data of the multi-source data acquisition module and generate a spatio-temporally aligned fusion data set by using an adaptive feature selection algorithm; The dual-channel time series modeling module: is used to receive the fusion data set and output a coupled feature vector through the parallel processing of the surface deformation feature extraction channel and the groundwater anomaly detection channel; The risk probability dynamic deduction module: is used to input the coupled feature vector into an LSTM-Transformer hybrid model optimized by transfer learning to generate a regional geological risk probability distribution map; The multi-dimensional coupling evaluation module: is used to generate an evaluation matrix including a stability index, interference sensitivity, and catastrophe diffusion rate according to the risk probability distribution map in combination with a geological structure feature library and a human activity database; The three-dimensional visualization warning module: is used to receive the evaluation matrix data, generate an interactive risk map through dynamic isosurface rendering technology, and trigger a hierarchical warning signal.

2. The geological disaster risk assessment platform based on big data according to claim 1, wherein, The multi-source data acquisition module includes a satellite remote sensing data acquisition unit, a groundwater level monitoring data acquisition unit, a surface displacement sensor data acquisition unit, and a meteorological data acquisition unit; among them: The satellite remote sensing data acquisition unit: collects spectral data of the ground area in real time through remote sensing sensors deployed on satellites, including data in visible light, infrared, and microwave bands; The groundwater level monitoring data acquisition unit: monitors the change of the groundwater level in real time through a water level monitoring sensor installed in the groundwater source; The surface displacement sensor data acquisition unit: monitors the surface displacement in real time through a displacement sensor installed on the ground and records the direction and amplitude of the ground displacement; The meteorological data acquisition unit: is used to collect meteorological data in real time, including temperature, humidity, wind speed, and precipitation.

3. The geological disaster risk assessment platform based on big data according to claim 1, characterized in that The dynamic weight fusion module includes a data preprocessing unit, an adaptive feature selection unit, and a spatio-temporal alignment unit; among them: The data preprocessing unit: is used to preprocess the data output by the multi-source data acquisition module, including data denoising, normalization, and missing value filling, to ensure that the data formats of different data sources are consistent; The adaptive feature selection unit: analyzes the preprocessed data through an adaptive feature selection algorithm, automatically identifies and selects the most relevant feature variables, and excludes redundant information; The spatio-temporal alignment unit: is used to align the feature data from different data sources in terms of time and space to ensure the matching of timestamps and geographical locations of each data source, so as to generate a spatio-temporally aligned fusion data set.

4. The geological disaster risk assessment platform based on big data according to claim 3, characterized in that The adaptive feature selection unit includes: Feature importance calculation: uses the Pearson correlation coefficient to measure the linear relationship between each feature and the target variable; Redundant feature elimination: For the redundant features in the feature set, by calculating the correlation between features, those features that are highly correlated with other features are removed. Specifically, when the correlation between a certain feature and other features exceeds a certain predetermined threshold, the feature exceeding the predetermined threshold is considered a redundant feature and is thus eliminated.

5. The geological disaster risk assessment platform based on big data according to claim 1, characterized in that, The dual-channel time series modeling module includes a surface deformation feature extraction channel, a groundwater hydrological anomaly detection channel, and a coupled feature vector output unit; where: Surface deformation feature extraction channel: Used to receive surface displacement data from the multi-source data acquisition module and extract the features of surface deformation through time series modeling techniques. Groundwater hydrological anomaly detection channel: Used to receive groundwater level data and extract groundwater hydrological anomaly features through time series analysis techniques. Coupled feature vector output unit: Used to perform parallel processing on the output results of the surface deformation feature extraction channel and the groundwater hydrological anomaly detection channel, and fuse the feature vectors of the two channels in series to generate a comprehensive coupled feature vector.

6. The geological disaster risk assessment platform based on big data according to claim 5, wherein The coupled feature vector output unit includes: Feature vector normalization: Normalize the surface deformation feature vector and the groundwater hydrological anomaly feature vector. Let the surface deformation feature vector be F surface , and the groundwater hydrological anomaly feature vector be F water . Then the normalization formula is: and where μ surface and σ surface are the mean and standard deviation of the surface deformation feature vector respectively, μ water and σ water are the mean and standard deviation of the groundwater hydrological anomaly feature vector respectively, F surface ′ and F water ′ are the normalized feature vectors; Feature vector concatenation: The standardized surface deformation feature vector and the groundwater hydrological anomaly feature vector are fused through a concatenation operation. The concatenation operation arranges the two feature vectors in sequence to form a comprehensive coupled feature vector, denoted as: F coupled = [F surface '; F water '], where F coupled is the output comprehensive coupled feature vector, F surface ' and F water ' are the standardized surface deformation feature vector and the groundwater hydrological anomaly feature vector, and the symbol ; represents the vector concatenation operation.

7. The geological disaster risk assessment platform based on big data according to claim 6, wherein, The risk probability dynamic deduction module includes a coupled feature vector input unit, an LSTM unit, a Transformer unit, and a geological risk probability distribution map generation unit; where: Coupled feature vector input unit: used to receive the coupled feature vector F output from the dual-channel time series modeling module, and take it as input data to be passed to the LSTM-Transformer hybrid model for analysis; coupled , and use it as input data to be passed to the LSTM-Transformer hybrid model for analysis; LSTM cell: used to capture time series patterns in the coupled feature vectors, selectively preserve important historical information through the memory mechanism, and ignore irrelevant parts. Its calculation formula is: h t = σ(W h x t + b h ) · tanh(W c c t-1 + b c ), where h t is the hidden state at the current moment, x t is the input at the current moment, c t-1 is the cell state at the previous moment, W h and W c are weight matrices, b h and b c are bias terms; Transformer unit: used to capture long-term dependencies in the coupled feature vectors and enhance the parallel processing ability of the model. Transformer processes the relationships between various parts of the input data through the self-attention mechanism, capturing global information in the input data. The calculation formula of the self-attention mechanism is: where Q, K, and V are the query, key, and value respectively, and d k is the dimension of the key; Geological risk probability distribution map generation unit: Used to convert the output data processed by the LSTM-Transformer hybrid model into a geological risk probability distribution map.

8. The geological disaster risk assessment platform based on big data according to claim 1, wherein The multi-dimensional coupling evaluation module includes a geological structure feature extraction unit, a human activity impact evaluation unit, a catastrophe diffusion rate calculation unit, and an evaluation matrix generation unit; where: Geological structure feature extraction unit: used to extract relevant data from the geological structure feature library and calculate the geological stability of each region. The formula is: Among them, SI(x, y) is the stability index at position (x, y), and G i (x, y) is the value of the i-th geological feature at position (x, y), and w i is the weight of this geological feature, and n1 is the total number of geological features; Human activity impact assessment unit: used to extract activity data related to geological disaster risks from the human activity database, including population density, urbanization level, and infrastructure construction factors; and calculate the interference sensitivity based on these relevant activity data. The formula is: where IS(x, y) is the interference sensitivity at location (x, y), and A i (x, y) is the value of the i-th human activity at location (x, y), and α i is the impact factor of this human activity, and m is the number of human activity factors; Catastrophe diffusion rate calculation unit: used to calculate the catastrophe diffusion rate by analyzing the propagation mode of geological disasters in the area. The formula is: Where DR(x, y) is the catastrophe diffusion rate at position (x, y), P(x, y) is the geological disaster risk probability at this position, d(x, y) is the geological disaster propagation distance at this position, and k is the diffusion coefficient; Evaluation matrix generation unit: Used to integrate the data extracted from the geological structure feature library and the human activity database, as well as the stability index, interference sensitivity, and catastrophe diffusion rate calculated from the above units, to generate an evaluation matrix containing these three indicators.

9. The geological disaster risk assessment platform based on big data according to claim 1, characterized in that The three-dimensional visualization warning module includes a dynamic isosurface rendering unit, an interactive atlas generation unit, and a graded warning signal triggering unit; where: Dynamic isosurface rendering unit: Using dynamic isosurface rendering technology, generate a three-dimensional risk atlas based on the risk data in the evaluation matrix. Specifically, first determine the corresponding risk level according to the stability index, interference sensitivity, and catastrophe diffusion rate of each region, and map the risk level to different color or transparency values; then use the isosurface algorithm to generate an isosurface representing the degree of risk and display the risk distribution of different regions. Interactive atlas generation unit: Used to generate an interactive three-dimensional model through the rendered risk atlas, allowing users to dynamically view the risk distribution of different regions through rotation and scaling operations. Graded warning signal triggering unit: Used to trigger corresponding graded warning signals according to the preset risk levels. The basis for dividing the warning levels is expressed as: Where threshold2 and threshold3 are the preset low-risk and high-risk thresholds respectively, P(x,y) is the risk probability at this location, and Alert Level is the triggered warning level.

10. A geological disaster risk assessment method based on big data, implemented by the geological disaster risk assessment platform according to any one of claims 1-9, characterized in that, Including the following steps: S1: Real-time collect satellite remote sensing data, groundwater level monitoring data, surface displacement data, and meteorological data of the ground area; S2: Preprocess the collected various types of data, including data denoising, normalization, and missing value filling; S3: Analyze the processed data through an adaptive feature selection algorithm, identify and select the most relevant feature variables, and exclude redundant information; S4: Based on the feature data selected in S3, extract the time series features of surface deformation and groundwater hydrological anomalies, and calculate the coupled feature vectors; S5: Input the coupled feature vectors into the LSTM-Transformer hybrid model for time series analysis and prediction, and generate a probability distribution map of regional geological risks; S6: According to the geological risk probability distribution map, combined with geological structure features and human activity data, calculate the stability index, interference sensitivity, and catastrophe diffusion rate of each region, and generate an evaluation matrix; S7: Generate a three-dimensional risk map according to the evaluation matrix, use dynamic isosurface rendering technology to visualize the geological disaster risks of different regions, and trigger corresponding warning signals according to the risk levels.

Images