A multi-observation-point combined data processing system based on geohazard analysis

Abstract

This invention belongs to the field of geological disaster hazard data technology and discloses a multi-observation point combined data processing system based on geological disaster hazard analysis. It includes a multi-source sensing layer, a data transmission layer, a core data fusion and processing layer, an application service layer, and a slope unit division module. This invention collects integrated multi-source monitoring data through the air-ground-ground-well multi-source sensing layer, and achieves stable data transmission by combining a dual-mode communication module with an UAV-delivered integrated monitoring station. It solves the problem of data asynchrony caused by different multi-source heterogeneous data formats, sampling frequencies, and accuracies in the prior art. It uses a point, surface, and volume collaborative analysis model to verify abnormal data in three dimensions, and generates refined early warning information for specific slope units by combining a multi-index comprehensive early warning model. The information is then pushed to the disaster prevention personnel through a geological disaster digital twin cockpit, forming a closed-loop processing flow of data collection, transmission, fusion, verification, early warning, and push.

Description

Technical Field

[0001] This invention belongs to the field of geological disaster hazard data technology, specifically a multi-observation point combined data processing system based on geological disaster hazard analysis. Background Technology

[0002] Geological disasters such as landslides, debris flows, and ground collapses are characterized by their suddenness, destructiveness, and wide distribution. Currently, geological disaster monitoring has evolved from single-sensor monitoring to comprehensive monitoring using multiple methods and parameters.

[0003] However, in existing technologies, the data formats, sampling frequencies, and accuracies acquired by monitoring methods such as spaceborne InSAR interferometric synthetic aperture radar, ground-based GNSS global navigation satellite systems, crack gauges, and rain gauges are inconsistent, often leading to data asynchrony issues. For example, the "Data Processing Method, Device, Electronic Equipment, and Storage Medium" (patent number: ZL202510799696.7) developed by the Hydrogeology and Environmental Geology Survey Center of the China Geological Survey focuses on the problem of processing missing data in landslide monitoring data, achieving effective completion of missing data and displacement trend prediction. However, this method mainly targets data completion for single data types and has not yet solved the problem of spatiotemporal registration between multi-source heterogeneous sensors. Secondly, regarding the issue of collaborative integration, for example, the "Method for Determining Critical Rainfall Threshold for Landslide Prediction" (Patent No.: ZL202510799696.7) proposed by Researcher Wang Tao's team at the Institute of Geomechanics analyzes the spatial correlation characteristics between landslide density and rainfall parameters using machine learning models such as random forests, and establishes a comprehensive early warning model based on characteristic rainfall parameters. Although this method improves the accuracy of the rainfall threshold, it still falls into the category of fixed parameter models and fails to achieve dynamic adjustment of sensor weights. The "A Method for Predicting Construction Environment Sudden Change Risk Fields by Integrating Geological and Meteorological Data" (Publication No.: CN120911980A) applied for by State Grid Hebei Electric Power Co., Ltd. introduces a method based on... The spatiotemporal adaptive graph neural network based on the intention mechanism can dynamically calculate feature weights based on real-time environmental data, representing an innovative application of dynamic weighting algorithms in the field of construction risk prediction. However, this method is mainly aimed at construction scenarios and has not yet been applied to the collaborative analysis of multiple observation points for geological hazard hazards. Thirdly, for example, the "A dual control system for geological hazard hazard points + risk areas" (application number: CN202311835262.5) disclosed by Jinan Dingxin Patent & Trademark Agency mentions the use of ArcGIS platform for pre-division of slope units, but it mainly relies on human-computer interaction for improvement, with limited automation. The "A system considering spatial hazards" developed by the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences and other units... The "Spatial Landslide Disaster Assessment Method and System" (Patent No.: CN202210247770.0) couples the spatial hazard zoning of landslide disasters with rainfall threshold early warning. Although this method considers spatial hazard factors, the spatial units are still relatively macroscopic. Another method, "An Integrated Evaluation Method for Landslide Area-Individual Unit Hazard Early Warning" (Application No.: CN202310211090.8), uses machine learning algorithms to construct a landslide area hazard assessment model based on slope units and establishes a weighted model for individual unit hazard early warning applicable to each evaluation unit, realizing the combination of area and individual units. However, this method focuses on hazard assessment and does not form a closed loop with the dynamic fusion of multi-source monitoring data. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-observation point combined data processing system based on geological disaster hazard analysis to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A multi-observation point combined data processing system based on geological disaster hazard analysis includes a multi-source sensing layer, a data transmission layer, a core data fusion and processing layer, an application service layer, and a slope unit division module; The multi-source sensing layer includes an airborne monitoring module, a space-based monitoring module, a ground-based monitoring module, and a downhole monitoring module; The data transmission layer includes a dual-mode communication module and an integrated monitoring station delivered by drone. The monitoring station integrates a miniature GNSS, a seismograph, a camera, and a rain gauge. The dual-mode communication module supports 4G and 5G public network communication, as well as BeiDou short message service and satellite communication. The core data fusion and processing layer includes a geological disaster big data center with a hybrid cloud architecture and a three-level fusion processing unit; The application service layer includes a geological disaster digital twin cockpit, the input of which is connected to the output of the decision-level fusion unit; The input end of the slope unit division module is connected to the high-precision DEM data source in the space-based monitoring module, and the output end is connected to the decision-level fusion unit. The slope unit division module is based on 2-meter grid high-precision DEM data.

[0006] Preferably, the airborne monitoring module includes UAV-borne InSAR and LiDAR, the space-based monitoring module includes spaceborne InSAR and BeiDou and GNSS receivers, the ground monitoring module includes crack gauges, rain gauges, soil moisture sensors, microseismic monitors and slope radars, and the downhole monitoring module includes downhole tiltmeters and stress gauges.

[0007] Preferably, the three-level fusion processing unit includes: a data-level fusion unit, a feature-level fusion unit, and a decision-level fusion unit. The input terminal of the data-level fusion unit is connected to the output terminals of the airborne monitoring module, the space-based monitoring module, the ground monitoring module, and the downhole monitoring module, respectively. The input terminal of the feature-level fusion unit is connected to the output terminal of the data-level fusion unit, and the input terminal of the decision-level fusion unit is connected to the output terminal of the feature-level fusion unit.

[0008] Preferably, the dynamic weight optimization algorithm running in the decision-level fusion unit is a dynamic weight multi-source data fusion algorithm based on improved ant colony intelligent optimization; the improved ant colony algorithm introduces an adaptive pheromone evaporation factor and a deformation acceleration guiding factor, specifically implemented through the following formula: EQ1: ; EQ2: ; EQ3: ; EQ4: ; EQ5: ; EQ6: ; EQ7: ; in, The total number of sensors participating in the fusion; For the ᵢth sensor in The fusion weights at each moment satisfy ; For the ᵢth sensor in Monitoring values ​​at any given time; for The result of the fusion of moments; for Measured values ​​from precise leveling at all times; The length of the observation period; The objective function value represents the root mean square of the fusion error; for The pheromone concentration along the time path (i,j); It is an adaptive volatility factor; Basic volatility factor, range of values ∈[0.1,0.5]; This is an adjustment coefficient, with a range of values. ∈[0.01,0.1]; for Deformation rate at time; For the first Only one ant in this iteration is on the path ( The increase in pheromone released on the surface; The pheromone enhancement constant has a range of values. ∈[1,10]; For the first The objective function value of the weight combination corresponding to each ant; The acceleration guiding coefficient has a range of values. ∈[0.1,0.3]; This is the normalized deformation acceleration value; for Deformation acceleration at time t; This represents the maximum absolute value of deformation acceleration during the historical observation period; for At time k, the k-th ant starts from node Transfer to node The probability of; This is a heuristic factor, taken as the normalized value of the sensor signal-to-noise ratio; This is an adjustment factor, with a range of values. ∈[1,2]; This is the set of possible next nodes for the current ant; The algorithm's iteration termination condition is satisfied. ,in To determine the convergence accuracy threshold, take... , To determine the maximum number of iterations, take... .

[0009] Preferably, the 2-meter grid high-precision DEM data is automatically divided using an improved slope aspect uniformity region growing algorithm, specifically through the following steps: Step A1: Input DEM data and calculate the aspect value of each grid cell. ,in They are respectively Rate of change of elevation in direction; Step A2: Set the slope consistency threshold Set a minimum slope unit area threshold. ; Step A3: Select an unmarked grid cell as the seed point, and perform eight-neighborhood region growth centered on the seed point. If the difference between the slope aspect of the neighboring grid cells and the slope aspect of the seed point is significant... If so, they will be merged into the same slope unit; Step A4: Calculate the area of ​​the initial unit formed by growth. If the unit area... If so, it will be merged into the adjacent unit with the smallest difference in slope aspect. The merging target selection criterion is: ,in Each is the current unit and adjacent units The average slope aspect value; Step A5: Smooth the boundaries of the partitioning results using the majority filtering method. Output the slope element vector boundary.

[0010] 6. A method for analyzing geological hazard risks based on dynamic weighting and multi-observation-point collaboration, comprising the following steps: Step S1: Collect multi-source monitoring data integrating air, space, ground, and well through the multi-source sensing layer; Step S2: Transmit the collected monitoring data to the core data fusion and processing layer through the data transmission layer; Step S3: The core data fusion and processing layer performs three-level fusion processing on the received multi-source monitoring data; Step S4: Construct a point, surface, and volume collaborative analysis model. When an anomaly is detected by a certain monitoring method, data from other relevant monitoring points in the area will be automatically retrieved for three-dimensional verification. Step S5: Based on the slope unit division results and combined with the multi-index comprehensive early warning model, generate refined early warning information for specific slope units; Step S6: Output the early warning results through the application service layer and push them to the person responsible for disaster prevention.

[0011] Preferably, the three-level fusion process in step S3 specifically includes: Step S31: Data-level fusion. Kalman filtering is used to perform spatiotemporal registration and preliminary fusion of InSAR and GNSS data to eliminate atmospheric delay errors. The prediction and update process of Kalman filtering is as follows: EQ8: ; EQ9: ; EQ10: ; EQ11: ; EQ12: ; EQ13: ; Step S32: Feature-level fusion, extracting deformation feature parameters, including: Deformation rate; Deformation acceleration; tangent angle, where The initial creep rate; Rainfall intensity, among which This refers to cumulative rainfall. Crack width variation rate, where The width of the crack; Step S33: Decision-level fusion, run the dynamic weight algorithm based on improved ant colony intelligent optimization, converge to the globally optimal weight configuration, and output the fusion result; in, for The prior state estimate vector at time t. For displacement, For rate, For acceleration; for The posterior state estimate vector at time t; Here is the state transition matrix. , The sampling interval; To control the input matrix; To control the input vector; Let be the process noise vector, which follows the... , The process noise covariance matrix; for The observation vector at time; The observation matrix; To observe the noise vector, it follows the following... , To observe the noise covariance matrix; To estimate the covariance matrix a priori; To estimate the covariance matrix for the posterior time; The Kalman gain matrix; It is an identity matrix.

[0012] Preferably, the point, surface, and volume collaborative analysis model in step S4 is as follows: When the slope radar detects an overall deformation trend in a certain area, the system automatically retrieves detailed data of GNSS points in that area and the rate of change of the crack gauge, forming a three-dimensional verification chain: the area radar detects the anomaly, the point GNSS pinpoints the exact location, and the crack gauge verifies the nature of the deformation. When the deformation trends of the three are consistent and all exceed the threshold, the disaster risk is confirmed. When the trends are contradictory, it is marked as a suspected anomaly and is continuously monitored.

[0013] Preferably, the multi-indicator comprehensive early warning model in step S5 includes: For landslide hazards, real-time tracking of tangent angle α, cumulative displacement acceleration a, and rainfall intensity is crucial. With crack width variation rate Define the normalization factor for each indicator: ; ,in The acceleration threshold; ,in The threshold for rainfall intensity; ,in The threshold for crack change rate; ,in .

[0014] Preferably, the warning level determination in step S5 is as follows: No warning ; Yellow Alert ; Orange alert ; Red Alert .

[0015] The beneficial effects of this invention are as follows: 1. This invention collects integrated multi-source monitoring data through a multi-source sensing layer covering air, ground, and well, and achieves stable data transmission by combining a dual-mode communication module with an integrated monitoring station delivered by UAV. This solves the problem of data asynchrony caused by different multi-source heterogeneous data formats, sampling frequencies, and accuracies in the prior art.

[0016] 2. Through the three-level fusion processing unit of the core data fusion and processing layer, Kalman filtering is used for data-level spatiotemporal registration and preliminary fusion. An improved ant colony intelligent optimization algorithm with adaptive pheromone evaporation factor and deformation acceleration guiding factor is introduced to realize decision-level dynamic weight fusion. This overcomes the defect of fixed parameter model in the existing technology that cannot dynamically adjust sensor weights, and improves the accuracy and adaptability of multi-source data fusion.

[0017] 3. This invention achieves fully automated and accurate division of slope units by using an improved slope aspect uniformity region growth algorithm based on 2-meter grid high-precision DEM data. This solves the problem that slope unit division in the prior art relies on human-computer interaction or macroscopic spatial units, and provides a foundation for refined geological disaster analysis.

[0018] 4. By using a point, surface, and volume collaborative analysis model to verify abnormal data in three dimensions, and combining it with a multi-indicator comprehensive early warning model to generate refined early warning information for specific slope units, the information is pushed to the disaster prevention personnel through the geological disaster digital twin cockpit. This forms a closed-loop processing flow of data collection, transmission, fusion, verification, early warning, and push, which solves the pain points of existing technologies such as the lack of multi-dimensional verification and disconnection from spatial units in early warning, reduces the false alarm rate of early warning, and improves the accuracy and timeliness of geological disaster hazard early warning. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the overall system architecture of the present invention; Figure 2 This is a flowchart of the dynamic weighting algorithm based on ant colony intelligence optimization of the present invention; Figure 3 This is a schematic diagram of the point, surface, and volume collaborative analysis model of the present invention; Figure 4 This is a flowchart illustrating the overall steps of the geological disaster hazard analysis method of the present invention. Detailed Implementation

[0020] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 like Figures 1 to 4 As shown, this embodiment of the invention provides a multi-observation point combined data processing system based on geological disaster hazard analysis, including a multi-source sensing layer, a data transmission layer, a core data fusion and processing layer, an application service layer, and a slope unit division module; The multi-source sensing layer includes an airborne monitoring module, a space-based monitoring module, a ground-based monitoring module, and a downhole monitoring module; The data transmission layer includes a dual-mode communication module and an integrated monitoring station delivered by drone. The monitoring station integrates a miniature GNSS, a seismograph, a camera, and a rain gauge. The dual-mode communication module supports 4G and 5G public network communication, as well as BeiDou short message service and satellite communication. The core data fusion and processing layer includes a geological disaster big data center with a hybrid cloud architecture and a three-level fusion processing unit; The application service layer includes a geological disaster digital twin cockpit, the input of which is connected to the output of the decision-level fusion unit; The input end of the slope unit division module is connected to the high-precision DEM data source in the space-based monitoring module, and the output end is connected to the decision-level fusion unit. The slope unit division module is based on 2-meter grid high-precision DEM data.

[0022] The airborne monitoring module includes UAV-borne InSAR and LiDAR, the space-based monitoring module includes spaceborne InSAR and BeiDou and GNSS receivers, the ground monitoring module includes crack gauges, rain gauges, soil moisture sensors, microseismic monitoring instruments and slope radar, and the downhole monitoring module includes downhole tiltmeters and stress gauges.

[0023] The three-level fusion processing unit includes a data-level fusion unit, a feature-level fusion unit, and a decision-level fusion unit. The input of the data-level fusion unit is connected to the output of the airborne monitoring module, the space-based monitoring module, the ground monitoring module, and the downhole monitoring module, respectively. The input of the feature-level fusion unit is connected to the output of the data-level fusion unit, and the input of the decision-level fusion unit is connected to the output of the feature-level fusion unit.

[0024] The dynamic weight optimization algorithm running in the decision-level fusion unit is a dynamic weight multi-source data fusion algorithm based on improved ant colony intelligent optimization. The improved ant colony algorithm introduces an adaptive pheromone evaporation factor and a deformation acceleration guiding factor, which are specifically implemented through the following formulas: EQ1: ; EQ2: ; EQ3: ; EQ4: ; EQ5: ; EQ6: ; EQ7: ; in, The total number of sensors participating in the fusion; For the ᵢth sensor in The fusion weights at each moment satisfy ; For the ᵢth sensor in Monitoring values ​​at any given time; for The result of the fusion of moments; for Measured values ​​from precise leveling at all times; The length of the observation period; The objective function value represents the root mean square of the fusion error; for The pheromone concentration along the time path (i,j); It is an adaptive volatility factor; Basic volatility factor, range of values ∈[0.1,0.5]; This is an adjustment coefficient, with a range of values. ∈[0.01,0.1]; for Deformation rate at time; For the first Only one ant in this iteration is on the path ( The increase in pheromone released on the surface; The pheromone enhancement constant has a range of values. ∈[1,10]; For the first The objective function value of the weight combination corresponding to each ant; The acceleration guiding coefficient has a range of values. ∈[0.1,0.3]; This is the normalized deformation acceleration value; for Deformation acceleration at time t; This represents the maximum absolute value of deformation acceleration during the historical observation period; for At time k, the k-th ant starts from node Transfer to node The probability of; This is a heuristic factor, taken as the normalized value of the sensor signal-to-noise ratio; This is an adjustment factor, with a range of values. ∈[1,2]; This is the set of possible next nodes for the current ant; The algorithm's iteration termination condition is satisfied. ,in To determine the convergence accuracy threshold, take... , To determine the maximum number of iterations, take... .

[0025] The 2-meter grid high-precision DEM data is automatically divided using an improved slope aspect uniformity region growing algorithm, specifically through the following steps: Step A1: Input DEM data and calculate the aspect value of each grid cell. ,in They are respectively Rate of change of elevation in direction; Step A2: Set the slope consistency threshold Set a minimum slope unit area threshold. ; Step A3: Select an unmarked grid cell as the seed point, and perform eight-neighborhood region growth centered on the seed point. If the difference between the slope aspect of the neighboring grid cells and the slope aspect of the seed point is significant... If so, they will be merged into the same slope unit; Step A4: Calculate the area of ​​the initial unit formed by growth. If the unit area... If so, it will be merged into the adjacent unit with the smallest difference in slope aspect. The merging target selection criterion is: ,in Each is the current unit and adjacent units The average slope aspect value; Step A5: Smooth the boundaries of the partitioning results using the majority filtering method. Output the slope element vector boundary; Each slope element is used as the smallest independent unit for geological hazard analysis, and the element attribute vector is represented as follows: ,in k is a unique identifier for the unit. The average slope aspect of the unit. The average slope of the unit. For unit area, The perimeter of the unit.

[0026] Example 2 A method for analyzing geological hazard risks based on dynamic weighting and multi-observation-point collaboration includes the following steps: Step S1: Collect multi-source monitoring data integrating air, space, ground, and well through the multi-source sensing layer; Step S2: Transmit the collected monitoring data to the core data fusion and processing layer through the data transmission layer; Step S3: The core data fusion and processing layer performs three-level fusion processing on the received multi-source monitoring data; Step S4: Construct a point, surface, and volume collaborative analysis model. When an anomaly is detected by a certain monitoring method, data from other relevant monitoring points in the area will be automatically retrieved for three-dimensional verification. Step S5: Based on the slope unit division results and combined with the multi-index comprehensive early warning model, generate refined early warning information for specific slope units; Step S6: Output the early warning results through the application service layer and push them to the person responsible for disaster prevention.

[0027] Specifically, the three-level fusion process in step S3 includes: Step S31: Data-level fusion. Kalman filtering is used to perform spatiotemporal registration and preliminary fusion of InSAR and GNSS data to eliminate atmospheric delay errors. The prediction and update process of Kalman filtering is as follows: EQ8: ; EQ9: ; EQ10: ; EQ11: ; EQ12: ; EQ13: ; Step S32: Feature-level fusion, extracting deformation feature parameters, including: Deformation rate; Deformation acceleration; tangent angle, where The initial creep rate; Rainfall intensity, among which This refers to cumulative rainfall. Crack width variation rate, where The width of the crack; Step S33: Decision-level fusion, run the dynamic weight algorithm based on improved ant colony intelligent optimization, converge to the globally optimal weight configuration, and output the fusion result; in, for The prior state estimate vector at time t. For displacement, For rate, For acceleration; for The posterior state estimate vector at time t; Here is the state transition matrix. , The sampling interval; To control the input matrix; To control the input vector; Let be the process noise vector, which follows the... , The process noise covariance matrix; for The observation vector at time; The observation matrix; To observe the noise vector, it follows the following... , To observe the noise covariance matrix; To estimate the covariance matrix a priori; To estimate the covariance matrix for the posterior time; The Kalman gain matrix; It is an identity matrix.

[0028] Specifically, the point, surface, and volume collaborative analysis model in step S4 is as follows: When the slope radar detects an overall deformation trend in a certain area, the system automatically retrieves detailed data of GNSS points in that area and the rate of change of the crack gauge, forming a three-dimensional verification chain: the area radar detects the anomaly, the point GNSS pinpoints the exact location, and the crack gauge verifies the nature of the deformation. When the deformation trends of the three are consistent and all exceed the threshold, the disaster risk is confirmed. When the trends are contradictory, it is marked as a suspected anomaly and is continuously monitored.

[0029] The multi-indicator comprehensive early warning model in step S5 includes: The multi-indicator comprehensive early warning model in step S5 includes: For landslide hazards, real-time tracking of tangent angle α, cumulative displacement acceleration a, and rainfall intensity is crucial. With crack width variation rate Define the normalization factor for each indicator: ; ,in The acceleration threshold; ,in The threshold for rainfall intensity; ,in The threshold for crack change rate; ,in .

[0030] The warning level determination in step S5 is as follows: No warning ; Yellow Alert ; Orange alert ; Red Alert ; By analyzing terraced loess landslides, we can identify their convergent displacement-time curves. ,in For the final convergence displacement, Here are the convergence coefficients; the stability criterion is... And the duration is continuously satisfied. Hours, of which As the rate threshold, take / sky; For flash floods and debris flows, integrate minute-level data from rain gauges. and microseismic monitoring data The warning trigger condition is and And duration ,in The rainfall intensity threshold is taken as... mm / hour The vibration amplitude threshold is set to 0.1 m / s². The duration threshold is set to 30 minutes, and the advance warning time is calculated as follows: , Among them To estimate the travel time of debris flows from their formation area to the protected area, This represents the time elapsed since the formation zone began.

[0031] The operating principle of this system is as follows: This system utilizes a multi-source sensing layer with airborne, space-based, ground-based, and underground monitoring modules to collect multi-dimensional geological disaster monitoring data, including regional surface deformation, topography, large-scale surface displacement, spatial location, local surface deformation, hydrology, soil, geological activity parameters, and underground rock mass stress changes and displacement characteristics. The data transmission layer's dual-mode communication module supports 4G, 5G public network communication, BeiDou short message service, and satellite communication, adapting to different communication needs. A drone-delivered integrated monitoring station can be quickly deployed to remote locations to complete data collection, storage, and transmission, ensuring continuous and comprehensive data transmission. The core data fusion and processing layer's three-level fusion processing unit progressively fuses data at the data level, feature level, and decision level. At the data level, Kalman filtering eliminates errors and achieves spatiotemporal registration. At the feature level, core parameters such as deformation rate and deformation acceleration are extracted. For decision-making... The system employs an improved ant colony intelligent optimization algorithm to iteratively solve for the optimal sensor weight configuration and output fusion results. The slope unit partitioning module, based on 2-meter grid high-precision DEM data, performs slope aspect calculation, eight-neighbor generation, area screening, and boundary smoothing steps to output slope unit vector boundaries with attribute vectors, serving as the smallest independent spatial unit for geological disaster analysis. The geological disaster hazard analysis operates according to the steps of data acquisition and transmission, three-level fusion processing, point, surface, and volume collaborative verification, refined early warning generation, and early warning result push. When an anomaly is detected by a certain monitoring method, relevant data is retrieved for verification, and the early warning level is determined by combining a multi-indicator comprehensive early warning model. The early warning information is visualized and pushed through a geological disaster digital twin cockpit. The system also analyzes disaster causes based on historical and new data, automatically optimizes algorithm parameters, and continuously improves fusion accuracy and early warning precision, forming a fully closed-loop geological disaster hazard handling and early warning system.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-observation-point combined data processing system based on geological disaster hazard analysis, characterized in that: It includes a multi-source sensing layer, a data transmission layer, a core data fusion and processing layer, an application service layer, and a ramp unit partitioning module; The multi-source sensing layer includes an airborne monitoring module, a space-based monitoring module, a ground-based monitoring module, and a downhole monitoring module; The data transmission layer includes a dual-mode communication module and an integrated monitoring station delivered by drone. The monitoring station integrates a miniature GNSS, a seismograph, a camera, and a rain gauge. The dual-mode communication module supports 4G and 5G public network communication, as well as BeiDou short message service and satellite communication. The core data fusion and processing layer includes a geological disaster big data center with a hybrid cloud architecture and a three-level fusion processing unit; The application service layer includes a digital twin cockpit for geological disasters, the input of which is connected to the output of the decision-level fusion unit; The input end of the slope unit division module is connected to the high-precision DEM data source in the space-based monitoring module, and the output end is connected to the decision-level fusion unit. The slope unit division module is based on 2-meter grid high-precision DEM data.

2. The multi-observation point combined data processing system based on geological disaster hazard analysis according to claim 1, characterized in that: The airborne monitoring module includes UAV-borne InSAR and LiDAR, the space-based monitoring module includes satellite-borne InSAR and BeiDou and GNSS receivers, the ground monitoring module includes crack gauges, rain gauges, soil moisture sensors, microseismic monitors and slope radars, and the downhole monitoring module includes downhole tiltmeters and stress gauges.

3. The multi-observation point combined data processing system based on geological disaster hazard analysis according to claim 1, characterized in that: The three-level fusion processing unit includes a data-level fusion unit, a feature-level fusion unit, and a decision-level fusion unit. The input terminal of the data-level fusion unit is connected to the output terminals of the airborne monitoring module, the space-based monitoring module, the ground monitoring module, and the downhole monitoring module, respectively. The input terminal of the feature-level fusion unit is connected to the output terminal of the data-level fusion unit, and the input terminal of the decision-level fusion unit is connected to the output terminal of the feature-level fusion unit.

4. The multi-observation point combined data processing system based on geological disaster hazard analysis according to claim 1, characterized in that: The dynamic weight optimization algorithm running in the decision-level fusion unit is a dynamic weight multi-source data fusion algorithm based on improved ant colony intelligent optimization; the improved ant colony algorithm introduces an adaptive pheromone evaporation factor and a deformation acceleration guiding factor, specifically implemented through the following formulas: EQ1: ; EQ2: ; EQ3: ; EQ4: ; EQ5: ; EQ6: ; EQ7: ; in, The total number of sensors participating in the fusion; For the ᵢth sensor in The fusion weights at each moment satisfy ; For the ᵢth sensor in Monitoring values ​​at any given time; for The result of the fusion of moments; for Measured values ​​from precise leveling at all times; The length of the observation period; The objective function value represents the root mean square of the fusion error; for The pheromone concentration along the time path (i,j); It is an adaptive volatility factor; Basic volatility factor, range of values ∈[0.1,0.5]; This is an adjustment coefficient, with a range of values. ∈[0.01,0.1]; for Deformation rate at time; For the first Only one ant in this iteration is on the path ( The increase in pheromone released on the surface; The pheromone enhancement constant has a range of values. ∈[1,10]; For the first The objective function value of the weight combination corresponding to each ant; The acceleration guiding coefficient has a range of values. ∈[0.1,0.3]; This is the normalized deformation acceleration value; for Deformation acceleration at time t; This represents the maximum absolute value of deformation acceleration during the historical observation period; for At time k, the k-th ant starts from node Transfer to node The probability of; This is a heuristic factor, taken as the normalized value of the sensor signal-to-noise ratio; This is an adjustment factor, with a range of values. ∈[1,2]; This is the set of next nodes that the current ant can choose; The algorithm's iteration termination condition is satisfied. ,in To determine the convergence accuracy threshold, take... , To determine the maximum number of iterations, take... .

5. A multi-observation-point combined data processing system based on geological disaster hazard analysis according to claim 1, characterized in that: The 2-meter grid high-precision DEM data is automatically divided using an improved slope aspect uniformity region growing algorithm, specifically through the following steps: Step A1: Input DEM data and calculate the aspect value of each grid cell. ,in They are respectively Rate of change of elevation in direction; Step A2: Set the slope consistency threshold Set a minimum slope unit area threshold. ; Step A3: Select an unmarked grid cell as the seed point, and perform eight-neighborhood region growth centered on the seed point. If the difference between the slope aspect of the neighboring grid cells and the slope aspect of the seed point is significant... If so, they will be merged into the same slope unit; Step A4: Calculate the area of ​​the initial unit formed by growth. If the unit area... If so, it will be merged into the adjacent unit with the smallest difference in slope aspect. The merging target selection criterion is: ,in Each is the current unit and adjacent units The average slope aspect value; Step A5: Smooth the boundaries of the partitioning results using the majority filtering method. Output the slope element vector boundary.

6. A method for analyzing geological disaster risks based on dynamic weighting and multi-observation-point collaboration, implemented based on the system described in any one of claims 1-5, characterized in that, Includes the following steps: Step S1: Collect multi-source monitoring data integrating air, space, ground, and well through the multi-source sensing layer; Step S2: Transmit the collected monitoring data to the core data fusion and processing layer through the data transmission layer; Step S3: The core data fusion and processing layer performs three-level fusion processing on the received multi-source monitoring data; Step S4: Construct a point, surface, and volume collaborative analysis model. When an anomaly is detected by a certain monitoring method, data from other relevant monitoring points in the area will be automatically retrieved for three-dimensional verification. Step S5: Based on the slope unit division results and combined with the multi-index comprehensive early warning model, generate refined early warning information for specific slope units; Step S6: Output the early warning results through the application service layer and push them to the person responsible for disaster prevention.

7. The geological hazard analysis method based on dynamic weighting and multi-observation point collaboration according to claim 6, characterized in that: The three-level fusion process in step S3 specifically includes: Step S31: Data-level fusion. Kalman filtering is used to perform spatiotemporal registration and preliminary fusion of InSAR and GNSS data to eliminate atmospheric delay errors. The prediction and update process of Kalman filtering is as follows: EQ8: ; EQ9: ; EQ10: ; EQ11: ; EQ12: ; EQ13: ; Step S32: Feature-level fusion, extracting deformation feature parameters, including: Deformation rate; Deformation acceleration; tangent angle, where The initial creep rate; Rainfall intensity, among which This refers to cumulative rainfall. Crack width variation rate, where The width of the crack; Step S33: Decision-level fusion, run the dynamic weight algorithm based on improved ant colony intelligent optimization, converge to the globally optimal weight configuration, and output the fusion result; in, for The prior state estimate vector at time t. For displacement, For rate, For acceleration; for The posterior state estimate vector at time t; Here is the state transition matrix. , The sampling interval; To control the input matrix; To control the input vector; Let be the process noise vector, which follows the... , The process noise covariance matrix; for The observation vector at time; The observation matrix; To observe the noise vector, it follows the following... , To observe the noise covariance matrix; To estimate the covariance matrix a priori; To estimate the covariance matrix for the posterior time; The Kalman gain matrix; It is an identity matrix.

8. The geological hazard analysis method based on dynamic weighting and multi-observation point collaboration according to claim 6, characterized in that: The point, surface, and volume collaborative analysis model in step S4 is specifically as follows: When the slope radar detects an overall deformation trend in a certain area, the system automatically retrieves detailed data of GNSS points in that area and the rate of change of the crack gauge, forming a three-dimensional verification chain: the area radar detects the anomaly, the point GNSS pinpoints the exact location, and the crack gauge verifies the nature of the deformation. When the deformation trends of the three are consistent and all exceed the threshold, the disaster risk is confirmed. When the trends are contradictory, it is marked as a suspected anomaly and is continuously monitored.

9. The geological hazard analysis method based on dynamic weighting and multi-observation point collaboration according to claim 6, characterized in that: The multi-indicator comprehensive early warning model in step S5 includes: For landslide hazards, real-time tracking of tangent angle α, cumulative displacement acceleration a, and rainfall intensity is crucial. With crack width variation rate Define the normalization factor for each indicator: ； ,in The acceleration threshold; ,in The threshold for rainfall intensity; ,in The threshold for crack change rate; ,in 。 10. The geological hazard analysis method based on dynamic weighting and multi-observation point collaboration according to claim 9, characterized in that: The warning level determination in step S5 is as follows: No warning ; Yellow Alert ; Orange alert ; Red Alert .

Images